engine/tortoise: sentence chunking + device fix + pitch/rate modulation

Catches up engines/tortoise/server.py with what's been deployed on
Lucy through tonight's smoke iterations:

0.2 — _chunk_for_tortoise splits text nodes at sentence boundaries
      (max 220 chars) before each tts_with_preset call. Fixes the
      end-of-prompt gibberish past tortoise's ~20s reliable horizon.

0.3 — _get_voice now .to(DEVICE) cached samples + latents. Without
      this, non-lj voices crash with 'Expected all tensors to be on
      the same device, but found cpu and cuda:0'.

0.4 — [voice:NAME pitch=N rate=R][/voice] tag syntax. librosa
      pitch_shift + time_stretch applied per-chunk for single-voice
      multi-character renders. The strategy survived the design
      table — but the librosa phase-vocoder artifacts at ±5 semitones
      ate the quality on the 2070 Super. Parked here for the GPU
      rebuild; modulation works architecturally, just needs better
      stretching algorithm (rubberband) + more headroom.

Production stayed Kokoro. Coast-Down preferred_voice_id reverted
to kokoro_af_heart in the live DB after this experiment.

.forgejo/workflows/gitleaks.yml

@@ -1,40 +0,0 @@
-# .forgejo/workflows/gitleaks.yml
-#
-# Sulkta canonical gitleaks workflow. Drop a copy into every public repo at
-# `.forgejo/workflows/gitleaks.yml` after the Forgejo act_runner is registered
-# (task #295).
-#
-# Pairs with the pre-receive hook installed on every bare repo — that one is
-# the strict enforcement layer (rejects the push); this one provides the
-# per-PR red ✗ that branch-protection rules can require before merge.
-#
-# Layer 1 (this workflow): visible per-PR status, can be a required check.
-# Layer 2 (pre-receive hook): strict enforcement at the server.
-# Layer 3 (johnny5 cron sweep): nightly full-history sweep across all repos.
-
-name: gitleaks
-
-on:
-  push:
-  pull_request:
-
-jobs:
-  scan:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          # Full history — gitleaks needs depth to scan a commit range.
-          fetch-depth: 0
-
-      - name: install gitleaks
-        run: |
-          curl -sSL -o gl.tar.gz \
-            https://github.com/gitleaks/gitleaks/releases/download/v8.21.2/gitleaks_8.21.2_linux_x64.tar.gz
-          tar xzf gl.tar.gz gitleaks
-          chmod +x gitleaks
-          ./gitleaks version
-
-      - name: scan
-        run: |
-          ./gitleaks detect --source . --no-banner --redact --verbose

Cargo.toml

@@ -7,7 +7,7 @@ version = "0.0.1"
 edition = "2024"
 license = "MIT"
 authors = ["Cobb (Jacob Hayes)"]
-repository = "https://git.sulkta.com/cobb/skald"
+repository = "http://192.168.0.5:3001/cobb/skald"
 
 [workspace.dependencies]
 tokio = { version = "1", features = ["full"] }

Dockerfile

@@ -1,16 +1,26 @@
-# Multi-stage build for skald. Postgres ships in the same image
-# until the tool stabilises; to split, swap the runtime base to
-# debian:bookworm-slim, drop entrypoint.sh, point DATABASE_URL at
-# an external pg.
+# Multi-stage build for skald.
 #
-# Build context is the workspace root: `docker build -t skald:latest .`
+# Stage 1: compile the rust binary against rust:1-bookworm.
+# Stage 2: pgvector/pgvector:pg17 (debian-bookworm postgres with
+#          pgvector preinstalled) + tini + the skald binary.
+#
+# v0.1 ships postgres inside the same container ("singleton till we
+# have a real working tool"). When we extract the DB out, swap the
+# runtime base to debian:bookworm-slim, drop entrypoint.sh, point
+# DATABASE_URL at the external pg.
+#
+# Build context is the workspace root:
+#   docker build -t skald:latest .
 
+# ─── builder ──────────────────────────────────────────────────────
 FROM rust:1.95-bookworm AS builder
 WORKDIR /build
 
-# Dependency-cache layer: copy manifests + vendored path-dep first,
-# build stubs, then drop in real sources. clawdforge is a path dep
-# resolved at workspace load time.
+# Cache the dependency graph: copy manifests first, fetch + build
+# stubs, THEN drop in real sources. The vendored clawdforge SDK
+# needs its own manifest + source available during the cache layer
+# (path dep — Cargo resolves it at workspace load time, not at
+# crate-compile time).
 COPY Cargo.toml Cargo.lock ./
 COPY skald-core/Cargo.toml skald-core/Cargo.toml
 COPY skald/Cargo.toml      skald/Cargo.toml

LICENSE

@@ -1,21 +0,0 @@
-MIT License
-
-Copyright (c) 2026 Cobb (Jacob Hayes)
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.

README.md

@@ -1,79 +1,95 @@
 # skald
 
 Long-form story-writer with canon-keeping, sequel continuity, and
-self-hosted audiobook narration. The database is the source of truth;
-the binary is the tooling.
+(future) self-hosted audiobook narration. Database is the source of
+truth — the writer is the tooling.
 
-Named for the Old Norse poets who composed and memorized kings' sagas
-across generations.
+Named for the Old Norse poets who composed and memorized kings'
+sagas across generations.
 
-## What's wired
+## Status: v0.1 — scaffold
+
+What's wired:
 
 - Rust workspace (`skald-core` + `skald`)
 - Postgres schema for stories, characters, canon facts, chapters,
   passages, generation runs, audit findings, tags
-- pgvector for future similarity search
+- pgvector extension installed for future similarity search
 - `skald import-markdown` ingests a story file (chapters + bible)
-- `skald serve` exposes `/health` + the web inspector and runs
-  migrations on boot
-- `skald continue` runs gen → cleanup → audit per chapter, with
-  multi-chapter batching (cap 20)
-- `skald rewrite` re-authors a chapter in a named author's voice
-- `skald audit` runs whole-story prose-quality audit; `skald dedup`
-  is the surgical fix half of the loop
-- `skald prepare-narration` annotates a chapter with `[breath]` /
-  `[pause:Xs]` / `[scene]` beats and per-character `[voice:...]`
-  tags
-- `skald narrate` renders a chapter to audio via one of three TTS
-  engines (F5-TTS, Kokoro-82M, Tortoise-TTS) — see `engines/`
-- Named-author "soul" personas via `skald authors seed`; author
-  voice replaces the model's base system prompt for gen/cleanup/
-  rewrite/dedup/narrate_prep
+  into the schema
+- `skald serve` exposes `/health` and runs migrations on boot
+- Single-container deploy: postgres + skald in one image
+
+Wired (this commit):
+
+- clawdforge Rust SDK vendored at `vendor/clawdforge/` (upstream:
+  `Sulkta-Coop/clawdforge` `clients/rust/`)
+- `skald-core::forge` — three-pass orchestration shell (gen / cleanup /
+  audit). Prompts are TODO stubs; pipeline plumbing is in place.
+
+Not yet wired:
+
+- Web UI (the inbox + browse + queue surface)
+- Prompt templates for the three passes (heavy prompt-engineering
+  work — own session)
+- `skald-core::context` — assemble the LLM context blob from DB rows
+  (bible + characters + parent prose summaries + similarity-matched
+  passages)
+- Embeddings backfill + ivfflat index
+- TTS sidecar container + post-render audit chain (see
+  `docs/tts-pipeline.md`)
+
+## v0.1 smoke
+
+```bash
+docker compose -p skald up -d
+docker exec skald skald import-markdown \
+    --path /seed/coast-down.md \
+    --title "The Coast-Down"
+
+curl http://lucy:7780/health
+# → { ok: true, db_ok: true, story_count: 1, ... }
+```
 
 ## Schema (cheat sheet)
 
 ```
-stories             meta + status + parent/root for series
-authors             persona identity (slug, display_name, model)
-author_revisions    versioned souls; one current per author
-characters          real or fictional, story-scoped, voice-mappable
-canon_facts         setting, mystery, theme, rule, historical_anchor, hook
-chapters            full prose body + optional body_md_tts annotation
-chapter_summaries   short summaries for cheap context loading
-passages            paragraph-level + embedding vector(1536)
-voices              TTS voice rows (F5 ref clips / Kokoro / Tortoise names)
-pronunciation_overrides  per-story + global respellings for proper nouns
-generation_runs     every LLM call logged
-audit_findings      audit pass output (severity + area)
-narration_runs      per-chapter TTS renders
+stories         → meta + status + parent/root for series
+characters      → real or fictional, story-scoped
+canon_facts     → setting, mystery, theme, rule, historical_anchor, hook
+chapters        → full prose body
+chapter_summaries → short summaries for cheap context loading
+passages        → paragraph-level + embedding vector(1536)
+generation_runs → every LLM call logged
+audit_findings  → canon audit output (severity + area)
+tags            → arbitrary labels
 ```
 
-## Quickstart
+## Architecture (v0.1 + the plan)
 
-```sh
-docker compose up -d
-docker exec skald skald import-markdown --path /seed/<story>.md \
-    --title "<title>"
-curl http://localhost:7780/health
+```
+┌─────────────────────────────────┐
+│  skald container                │
+│  ┌───────────┐  ┌────────────┐  │
+│  │ postgres  │  │ skald-rust │  │
+│  │ pgvector  │←─│ axum + cli │  │
+│  │ localhost │  │ :7780      │  │
+│  └───────────┘  └─────┬──────┘  │
+└─────────────────────────┼────────┘
+                          │ HTTP (future)
+                          ↓
+                    ┌──────────┐
+                    │clawdforge│
+                    └─────┬────┘
+                          ↓
+                     opus calls
 ```
 
-The compose file expects `POSTGRES_PASSWORD` and (optionally)
-`CLAWDFORGE_URL` + `CLAWDFORGE_TOKEN` in `.env`. Story markdown
-goes into `./seed/`; postgres data persists in `./pgdata/`.
-
-## Architecture
-
-v0.1 ships postgres inside the skald container — singleton until
-the tool stabilises. To extract postgres later, swap the runtime
-base to `debian:bookworm-slim`, drop `entrypoint.sh`, and point
-`DATABASE_URL` at the external pg. The binary doesn't care where
-the DB lives.
-
-The generation passes call out to `clawdforge` (a bearer-token-gated
-HTTP wrapper around `claude -p`). The Rust client is vendored at
-`vendor/clawdforge/`. TTS calls go HTTP+JSON to the per-engine
-sidecars under `engines/`.
+v1.0+: extract postgres to its own container on db-net. skald
+becomes pure stateless rust, connects via `DATABASE_URL`. Migration
+is a connection-string change + a network move; the binary doesn't
+care where the DB lives.
 
 ## License
 
-MIT — see `LICENSE`.
+MIT.

compose.yml

@@ -1,42 +1,37 @@
-# Standalone compose stack for skald. Postgres lives inside the
-# skald container — single deployable unit until the tool stabilises.
+# Standalone compose stack for skald v0.1. Postgres lives in the
+# same container — single deployable unit "till we have a real
+# working tool" (cobb's call, 2026-05-13).
 #
-# Set in .env (or the environment):
-#   POSTGRES_PASSWORD=...           # required
-#   CLAWDFORGE_URL=http://...:8800  # if running gen / cleanup / audit
-#   CLAWDFORGE_TOKEN=cf_...
-#   SKALD_DATA=./pgdata             # optional override; defaults to ./pgdata
-#   SKALD_SEED=./seed               # optional override; defaults to ./seed
+# To deploy on Lucy:
+#   sudo mkdir -p /mnt/cache/appdata/skald/{pgdata,seed}
+#   sudo cp <story>.md /mnt/cache/appdata/skald/seed/
+#   sudo cp skald.env  /mnt/cache/appdata/secrets/skald.env  # POSTGRES_PASSWORD=...
+#   docker compose -p skald up -d
 #
 # To import the first story:
-#   docker compose exec skald skald import-markdown \
+#   docker exec skald skald import-markdown \
 #       --path /seed/<story>.md \
 #       --title "<title>"
-name: skald
 
 services:
   skald:
-    build: .
-    image: skald:latest
+    image: lucy-registry:5000/skald:latest
     container_name: skald
     restart: unless-stopped
     ports:
       - "7780:7780"
-    environment:
-      POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:?POSTGRES_PASSWORD is required}
-      POSTGRES_USER: ${POSTGRES_USER:-skald}
-      POSTGRES_DB: ${POSTGRES_DB:-skald}
-      DATABASE_URL: ${DATABASE_URL:-postgresql://skald:${POSTGRES_PASSWORD}@localhost:5432/skald}
-      CLAWDFORGE_URL: ${CLAWDFORGE_URL:-}
-      CLAWDFORGE_TOKEN: ${CLAWDFORGE_TOKEN:-}
-      SKALD_MODEL: ${SKALD_MODEL:-opus}
-      F5_TTS_URL: ${F5_TTS_URL:-}
-      KOKORO_URL: ${KOKORO_URL:-}
-      TORTOISE_URL: ${TORTOISE_URL:-}
-      RUST_LOG: ${RUST_LOG:-info}
-      SKALD_LOG_FORMAT: ${SKALD_LOG_FORMAT:-json}
+    env_file:
+      - /mnt/cache/appdata/secrets/skald.env
     volumes:
       # Postgres data — persist across container recreates.
-      - ${SKALD_DATA:-./pgdata}:/var/lib/postgresql/data
-      # Markdown corpus to import via `skald import-markdown`.
-      - ${SKALD_SEED:-./seed}:/seed:ro
+      - /mnt/cache/appdata/skald/pgdata:/var/lib/postgresql/data
+      # Markdown corpus to import via `docker exec skald skald import-markdown`.
+      - /mnt/cache/appdata/skald/seed:/seed:ro
+    environment:
+      RUST_LOG: ${RUST_LOG:-info}
+      SKALD_LOG_FORMAT: json
+    labels:
+      org.sulkta.domain: "sulkta"
+      org.sulkta.owner: "cobb"
+      org.sulkta.managed-by: "compose"
+      org.sulkta.role: "skald"

docs/authors.md

@@ -1,51 +1,83 @@
-# Authors as personas with souls
+# Authors as personas with souls (v0.3 design — IN PROGRESS)
 
-Each story has a named author with a soul. The author's voice bleeds
-through every generation pass — not as instruction stapled to the
-prompt, but as the substrate the prose grows from. The reader should
-feel "a person wrote this."
+The pivot from "anonymous generator" to "real-feeling human authors."
 
-Authors have memory across their corpus when the per-story
-`cross_story_memory` toggle is on. An author writing a Chernobyl
-piece can quietly echo a phrase from an earlier mining-strike story.
-Default is off — most stories stand alone.
+## Vision (cobb 2026-05-13)
 
-## Schema
+Each story has a **named author with a soul.** The author's voice
+bleeds through every pass — not as instruction stapled to the prompt,
+but as the substrate from which the prose grows. The reader should
+feel "a person wrote this." Example tagline cobb gave:
 
-Authors live in the database, not on disk. The soul is a markdown
-blob in `author_revisions.soul`. Authors are immutable; new soul
-revisions create a new `author_revisions` row marked `is_current`
-and the previous one is demoted.
+> "George Orwell but more rebel and pissed off — slightly comes out
+> in stories."
+
+Authors have **memory across their corpus.** Per-story opt-in toggle
+lets an author subtly reference / echo other stories they've
+written. "Cross-dimensional story crosses" — Orson Black writing a
+Chernobyl piece might quietly echo a phrase from his earlier
+mining-strike story when the toggle is on.
+
+## Why this is a real design change
+
+Until now skald is opus + a thin prompt. The output WILL eventually
+sound like opus-with-a-system-prompt. That's a ceiling on prose
+quality + identity. Authors-with-souls is the ceiling-lifter.
+
+Knock-on effects:
+- All three forge passes (gen / cleanup / audit) need author awareness.
+- Audit becomes nuanced: "does the prose match this author's voice?"
+  is an audit signal, not just "does it match canon?"
+- Cross-story memory unlocks long-arc literary projects (cobb's
+  potential YouTube channel idea, narrated-audiobook coop, etc).
+
+## Locked decisions (so far)
+
+1. **Authors live in the DB**, not on disk. Soul is a markdown blob
+   in `authors.soul`. Editable via web UI eventually; portable per
+   story.
+2. **Soul replaces opus's default system prompt** for every forge
+   pass on that story. clawdforge currently maps `system` to
+   `--append-system-prompt` — we'll either (a) live with appending,
+   or (b) extend clawdforge with a `system_mode: append|replace`
+   field. **Option (a) tonight** because clawdforge changes are
+   their own session.
+3. **Per-story toggle for cross-story memory** (`stories.cross_story_memory: bool`).
+   Per-author would make the toggle useless — we want Orson Black to
+   write standalone-Chernobyl AND deeply-cross-referenced-mining-strike.
+4. **Multiple authors per series allowed.** `stories.author_id` is
+   per-story, not per-root_story. Future-proofs collaborations.
+
+## Schema sketch
 
 ```sql
 CREATE TABLE authors (
-    id              UUID PRIMARY KEY,
-    slug            TEXT NOT NULL UNIQUE,
-    display_name    TEXT NOT NULL,
-    persona_tagline TEXT,
+    id              UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    slug            TEXT NOT NULL UNIQUE,            -- "orson-black"
+    display_name    TEXT NOT NULL,                   -- "Orson Black"
+    persona_tagline TEXT,                            -- "Orwell but more rebel + pissed off"
+    soul            TEXT NOT NULL,                   -- SOUL.md-style markdown blob
     model           TEXT NOT NULL DEFAULT 'opus',
-    is_synthetic    BOOLEAN NOT NULL DEFAULT true,
+    -- Optional per-author system-prompt scaffold. The author's soul
+    -- fills in the persona substance; the scaffold structures it
+    -- around skald's needs (canon honoring, no-meta-commentary,
+    -- etc). Default scaffold lives in skald-core.
+    system_template TEXT,
+    -- Tools the author can call during gen. Off by default — fiction
+    -- writes don't normally need WebSearch / Read. But an author
+    -- with research bent could opt in for fact-checking detours.
+    tools           TEXT[] NOT NULL DEFAULT '{}',
     created_at      TIMESTAMPTZ NOT NULL DEFAULT now(),
     updated_at      TIMESTAMPTZ NOT NULL DEFAULT now()
 );
 
-CREATE TABLE author_revisions (
-    id              UUID PRIMARY KEY,
-    author_id       UUID NOT NULL REFERENCES authors(id),
-    n               INT NOT NULL,
-    soul            TEXT NOT NULL,
-    system_template TEXT,
-    tools           TEXT[] NOT NULL DEFAULT '{}',
-    note            TEXT,
-    is_current      BOOLEAN NOT NULL DEFAULT false,
-    created_at      TIMESTAMPTZ NOT NULL DEFAULT now()
-);
-
 ALTER TABLE stories
-    ADD COLUMN author_id           UUID REFERENCES authors(id),
-    ADD COLUMN author_revision_id  UUID REFERENCES author_revisions(id),
+    ADD COLUMN author_id           UUID REFERENCES authors(id) ON DELETE SET NULL,
     ADD COLUMN cross_story_memory  BOOLEAN NOT NULL DEFAULT false;
 
+-- For cross_story_memory: which stories does this author have access
+-- to? Auto-row on every authored story; explicit "marked as read"
+-- rows for stories by other authors the author has internalized.
 CREATE TABLE author_corpus (
     author_id   UUID NOT NULL REFERENCES authors(id) ON DELETE CASCADE,
     story_id    UUID NOT NULL REFERENCES stories(id) ON DELETE CASCADE,
@@ -55,29 +87,11 @@ CREATE TABLE author_corpus (
 );
 ```
 
-## Per-pass author roles
+## Soul template (the contract per author)
 
-- **gen** — full author voice. They are writing.
-- **cleanup** — full author voice. Polishing their own draft, not a
-  neutral editor.
-- **rewrite** — full author voice. Re-authoring another hand's
-  prose; canon preserved, prose reworked.
-- **dedup** — full author voice. Surgical fix of audit-flagged
-  repetitions only.
-- **narrate_prep** — author voice if bound; the author's beat
-  placement carries.
-- **audit / prose_audit** — neutral. The audit checks the author's
-  work with fresh eyes; no author bound.
-- **summarize** — neutral. Continuity utility, not prose.
-
-When an author is bound, the soul replaces the model's base system
-prompt (`SystemMode::Replace`). Without an author, a neutral house
-scaffold is appended (`SystemMode::Append`).
-
-## Soul template
-
-Following the SOUL.md shape, recalibrated for authorial identity.
-Free-form prose under each section.
+Following the same shape as `/root/.openclaw/workspace/SOUL.md` but
+recalibrated for authorial identity. Suggested sections (free-form
+prose, not strict schema):
 
 ```markdown
 # Author: {{display_name}}
@@ -87,8 +101,8 @@ _Tagline: {{persona_tagline}}_
 ## Voice
 
 Sentence rhythm. Vocabulary register. Paragraph length tendencies.
-Dialogue density. Punctuation habits (dashes, semicolons, sentence
-fragments). What your prose SOUNDS like read aloud.
+Dialogue density. Punctuation habits (dashes, semicolons,
+sentence fragments). What your prose SOUNDS like read aloud.
 
 ## Worldview
 
@@ -98,13 +112,15 @@ in the implicit moral architecture of any scene.
 
 ## Specifics over abstractions
 
-The concrete details you reach for. The five senses you favor.
-Smells? Cold? Texture of cloth? Sound of machines?
+What concrete details you reach for. (Orwell: the cold tap, the
+miners' shoes, the gin.) The 5 senses you favor. Smells? Cold?
+Texture of cloth? Sound of machines?
 
 ## Pet peeves
 
 Words you refuse to write. Tropes you avoid. Sentimentalities you
-gut.
+gut. (Orson: "soul-stirring," "ineffable," "tapestry of
+emotions." No prose-poetry trim.)
 
 ## Sense of humor
 
@@ -113,51 +129,159 @@ does humor live — end of a sentence, mid-clause, or never?
 
 ## Biography (real or invented)
 
-A few biographical facts that EXPLAIN the voice — the formative
-cuts, not a CV.
+A few biographical facts that EXPLAIN the voice. (Not a CV — the
+formative cuts. Orson grew up in a coal town. His father died of
+black lung. He read Trotsky at seventeen. He spent two winters
+working in a Tyne shipyard.)
 
 ## Anchor authors
 
 Living or dead authors the prose draws from. Useful for the model
-to triangulate voice.
+to triangulate voice. (Orson: Orwell, Cormac McCarthy, Don
+DeLillo, Tony Judt's nonfiction.)
 
-## Do / Don't
+## Do
 
-Concrete prose moves to reach for and to avoid.
+- Specifics that puncture sentiment.
+- Direct address.
+- Cold air, hard surfaces.
+- Politics in the texture, not the lecture.
+
+## Don't
+
+- Soft consolations.
+- Therapy-speak.
+- Magical resolutions.
+- Adverbs that intensify ("absolutely," "incredibly," "deeply").
 ```
 
-The default scaffold (`DEFAULT_AUTHOR_SCAFFOLD` in `skald-core::forge`)
-wraps the soul in a system prompt that:
+The soul gets jammed into the system prompt via a scaffold like:
 
-- declares the model IS the author, not playing one
-- pins canon as non-negotiable (names, dates, established events)
-- forbids preamble, meta-commentary, fourth-wall breaks
-- substitutes the per-pass directive (`GEN_DIRECTIVE`,
-  `CLEANUP_DIRECTIVE`, `REWRITE_DIRECTIVE`, `DEDUP_DIRECTIVE`,
-  `NARRATE_PREP_DIRECTIVE`)
+```
+You are {{display_name}}, an author. Your voice and worldview are
+described in the soul below. Honor them in every sentence — not as
+performance but as substrate. The story's canon (characters,
+setting, established facts) is non-negotiable; your voice lives
+WITHIN those constraints.
 
-A per-author `system_template` overrides the default scaffold when
-set; otherwise the default is used.
+If your worldview makes the user-prompted plot uncomfortable,
+write through that discomfort — that's where good prose lives.
+Never break the fourth wall to comment on the task. No meta-prose.
 
-## Cross-story memory
+---
 
-When `stories.cross_story_memory = true`, the continuation context
-pulls characters / canon_facts / passages from every story the
-author has authored or marked-read, not just the parent chain.
+{{soul}}
 
-To keep token budget sane, cross-corpus pulls are summary-only by
-default. Embeddings-similarity (once wired) can surface direct
-callbacks.
+---
 
-## Seeding an author
-
-```sh
-skald authors seed \
-    --slug orson-black \
-    --display-name "Orson Black" \
-    --tagline "Orwell but more rebel and pissed off" \
-    --file seeds/authors/orson-black.md
+You are writing for an audience of one (your reader). The story
+matters. The voice matters. Both at once.
 ```
 
-This creates the author + first revision (or adds a new revision to
-an existing author, which becomes current).
+## Cross-story memory mechanics
+
+When `stories.cross_story_memory = true`:
+
+1. Continuation context pulls characters / canon_facts / passages
+   from EVERY story the author has authored or marked-read, not just
+   the parent chain.
+2. To keep token budget sane, cross-story pulls are **summary-only
+   by default** — full passages from other corpora are too token-
+   expensive and the model only needs flavor, not detail.
+3. Embeddings (when wired) can surface similarity-matched passages
+   from cross-corpus for direct callbacks.
+
+When `cross_story_memory = false` (default):
+- Context is just the parent chain. No cross-corpus pulls.
+- This is the right default — most stories should stand alone.
+
+## Per-pass author roles
+
+Authors carry voice through all three passes, but the lens shifts:
+
+- **gen**: full author voice. They're writing.
+- **cleanup**: full author voice. Polishing their own draft (not
+  a neutral editor).
+- **audit**: author SHIFTS to "canon auditor" mode — neutral, no
+  voice. The author isn't checking their own work; we want detached
+  canon-fact analysis. Schema-wise: audit ignores `stories.author_id`
+  and uses a default neutral system prompt.
+
+This is asymmetric in a useful way. The author writes + revises.
+The audit is the system's check on the author. Mirrors how books
+get edited in real life.
+
+## Seed authors (proposals — cobb picks)
+
+1. **Orson Black** — "Orwell but more rebel + pissed off." Cobb's
+   stated direction. Coast-Down sequel candidate.
+2. **Bay** — Bay's actual literary voice as captured in "Petal &
+   Bone" (saved in `memory/petal-and-bone.md` per PEOPLE.md). Soft,
+   observational, environmentally aware. For nature / family /
+   memory-shaped stories.
+3. **Kayos** (me) — ghost-mode, dry, technical, distrust-by-default.
+   Useful for cyberpunk-adjacent / hacker / industrial-decline
+   stories. Could write a parallel Chernobyl piece that focuses on
+   the dosimetry/control-room minutiae over the human lens.
+4. **House** — no soul, plain opus. For stories the user doesn't
+   want filtered through any specific personality. Like an
+   anonymous-author fallback.
+
+## Open questions (cobb decides)
+
+1. **Soul format** — strict section headings (template above) OR
+   free-form prose? The template helps the model know what to
+   prioritize; free-form lets each author breathe. *My lean:
+   strict headings for the seed authors, prove the shape works,
+   then loosen.*
+
+2. **Who's the first seed author for Coast-Down sequels?** Orson
+   Black makes thematic sense — the Soviet industrial-decline
+   subject matter. Bay would write a very different Coast-Down
+   sequel (more interior, more sensory, less political). Pick one
+   for the v0.3 demo or seed both?
+
+3. **Cross-story pull granularity** — summaries-only (v0.3 simple)
+   vs embeddings-similarity matched (needs embeddings wired first)
+   vs the full firehose (too expensive). *My lean: summaries-only
+   for v0.3, embeddings-similarity in v0.4 when we have multiple
+   stories per author.*
+
+4. **clawdforge `system_mode: replace`** — append now (today's
+   path), full replace later (clawdforge enhancement / future
+   Rust rewrite). Acceptable? Or worth doing the clawdforge change
+   first?
+
+5. **Audit pass author-neutrality** — should the author still
+   inform the audit at all? E.g., a "in this author's universe,
+   X is acceptable" could be a soul-derived note the audit
+   honors. *My lean: audit stays neutral. The author wrote the
+   thing; we want fresh eyes checking it.*
+
+6. **Author edit history** — does soul change over time matter?
+   Should we version souls so a story always knows "this was the
+   version of Orson Black active when I was written"? *My lean:
+   YES, easy v0.4 add — `authors.id` becomes immutable; a new
+   soul-revision creates a NEW author row with a `revision_of`
+   FK. Stories pin to a specific revision.*
+
+## Order of work after design lock
+
+1. Migration 0004: authors + stories.author_id + author_corpus + cross_story_memory.
+2. `skald-core::authors` module: load_author(slug), get_system_prompt_for_pass(author, kind).
+3. `skald-core::forge` updated: every pass takes a `&Author` (or None for "house"), builds system prompt from author + scaffold.
+4. `skald-core::context::ContinuationContext::assemble`: when story.cross_story_memory, pull from author_corpus instead of just parent chain.
+5. Seed the first author(s) via a `skald authors seed --persona orson-black` subcommand OR direct SQL bootstrap.
+6. THEN gen prompt template (the prose-craft session) lands AGAINST the authors layer.
+7. `skald continue --story <id>` wires it all together.
+
+## Status
+
+- [x] Brainstorm captured in this doc (2026-05-13 late)
+- [ ] Cobb's call on open questions 1-6
+- [ ] Schema migration 0004
+- [ ] authors module
+- [ ] forge author-aware
+- [ ] Seed Orson Black (?) soul
+- [ ] gen prompt template
+- [ ] CLI continue

engines/README.md

@@ -40,18 +40,19 @@ generalise. Examples:
   all three at once), preset choice ergonomics, character→tortoise-
   voice seed assignments.
 
-To deploy a tuned engine, check out the engine's branch in the build
-dir and `docker compose up -d --build`:
+When deploying an engine to Lucy, the build dir at
+`/mnt/cache/appdata/<engine>/build/` tracks the engine's branch:
 
 ```bash
+cd /mnt/cache/appdata/kokoro/build
 git fetch && git checkout engine/kokoro
-docker compose up -d --build
+docker compose -p <name> up -d --build
 ```
 
-## GPU coordination
+## GPU coordination (2070 Super)
 
-On an 8GB card F5 + Kokoro can co-reside (~5GB + ~1GB). Tortoise
-pushes the budget over and needs the GPU largely to itself — the
-`engine/tortoise` branch carries a script to stop kokoro + f5
-before a Tortoise run and restart them after. Replace with proper
-coordination once more VRAM is available.
+The 8GB card is the bottleneck. F5 + Kokoro can co-reside (~5GB +
+~1GB). Tortoise pushes the budget over and needs the GPU largely
+to itself — the `engine/tortoise` branch will carry the script
+that stops kokoro + f5 before a tortoise run and restarts them
+after. Replace with proper coordination once we have more VRAM.

engines/f5-tts/Dockerfile

@@ -1,6 +1,9 @@
-# F5-TTS rebuild on a known-good pytorch base. Upstream
-# ghcr.io/swivid/f5-tts:main shipped a torch/torchaudio ABI mismatch
-# that broke `import torchaudio` at boot; this image bypasses that.
+# Sulkta build of F5-TTS — upstream ghcr.io/swivid/f5-tts:main was
+# shipped with torch 2.11/torchaudio 2.4 ABI mismatch on 2026-05-13,
+# breaking import torchaudio at boot. We rebuild on a known-good
+# pytorch base + pip install f5-tts.
+#
+# Image tag in lucy-registry: lucy-registry:5000/f5-tts:<ver>
 #
 # License: Apache 2.0 (code) / CC-BY-NC (Emilia-trained weights).
 # Personal use OK; redistribution gray-area — flagged.
@@ -27,11 +30,12 @@ RUN pip install --no-cache-dir 'f5-tts>=1.0.0'
 # Pre-warm the HF cache directory.
 RUN mkdir -p /cache/hf /audio /voices
 
-COPY server.py /app/server.py
+COPY f5_server.py /app/f5_server.py
 WORKDIR /app
 
 EXPOSE 7860
 
-# Purpose-built FastAPI server, not Gradio. Models load at startup
-# so the first request doesn't pay the cold-start cost.
-CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860"]
+# Skald talks to our purpose-built FastAPI server, not Gradio.
+# Models load at startup (first request would otherwise pay the
+# cold-start cost). uvicorn on :7860 to keep the port stable.
+CMD ["uvicorn", "f5_server:app", "--host", "0.0.0.0", "--port", "7860"]

engines/f5-tts/compose.yml

@@ -1,21 +1,19 @@
-# F5-TTS sidecar.
+# F5-TTS standalone stack on Lucy.
 #
-# Code is Apache 2.0; pretrained F5TTS_v1_Base weights are CC-BY-NC
-# (Emilia training data). Personal use is fine; redistribution is a
-# flagged gray area.
+# License posture (acknowledged 2026-05-13): code is Apache 2.0, but
+# the pretrained model weights are CC-BY-NC (Emilia training data).
+# Personal listen is fine; public sharing is a flagged gray area.
+# Cobb's call: ship anyway.
 #
-# First run downloads ~2GB of model weights from HuggingFace into
-# the hf-cache volume; subsequent runs are warm.
+# Runtime: 8GB GPU is plenty (F5 inference ~4-6GB peak).
 #
-# Set in .env (or override):
-#   F5_HOST_PORT=7792
-#   F5_DATA=./data           # ${F5_DATA}/hf-cache + voices + audio
+# First-run cost: ~2GB model download from HuggingFace into hf-cache,
+# happens on first inference request. Subsequent runs are warm.
 name: f5-tts
 
 services:
   f5-tts:
-    build: .
-    image: f5-tts:0.3
+    image: lucy-registry:5000/f5-tts:0.3
     container_name: f5-tts
     restart: unless-stopped
     deploy:
@@ -26,11 +24,20 @@ services:
               count: all
               capabilities: [gpu]
     ports:
-      - "${F5_HOST_PORT:-7792}:7860"
+      - "192.168.0.5:7792:7860"
+      - "127.0.0.1:7792:7860"
     volumes:
-      - ${F5_DATA:-./data}/hf-cache:/cache/hf
-      - ${F5_DATA:-./data}/voices:/voices:ro
-      - ${F5_DATA:-./data}/audio:/audio
+      # HF model weights cache — persists ~2GB after first download.
+      - /mnt/cache/appdata/f5-tts/hf-cache:/cache/hf
+      # Reference voice clips (lj_speech.wav, etc).
+      - /mnt/cache/appdata/f5-tts/voices:/voices:ro
+      # Rendered audio output — skald writes story narrations here.
+      - /mnt/cache/appdata/f5-tts/audio:/audio
     environment:
       HF_HOME: /cache/hf
       HF_HUB_DISABLE_TELEMETRY: "1"
+    labels:
+      org.sulkta.domain: "sulkta"
+      org.sulkta.owner: "cobb"
+      org.sulkta.managed-by: "compose"
+      org.sulkta.role: "f5-tts"

engines/kokoro/Dockerfile

@@ -1,8 +1,13 @@
-# Kokoro-82M TTS. Apache 2.0 code AND weights — clean stack vs
-# F5-TTS's CC-BY-NC asterisk.
+# Sulkta build of Kokoro-82M TTS.
 #
-# Kokoro runs fine on CPU but we use the cuda base to stay
-# consistent with f5-tts and pick up the GPU when free.
+# License: Apache 2.0 (code AND model weights). Clean stack — no
+# CC-BY-NC asterisk like F5-TTS's Emilia weights. This is the
+# narrator engine for sleep-quality audiobook reads; F5-TTS stays
+# around for voice-cloning cases.
+#
+# Kokoro is small enough to run on CPU but we use the cuda base
+# anyway to stay consistent with f5-tts and so it'll pick up the
+# GPU when no other tenant has it.
 FROM pytorch/pytorch:2.6.0-cuda12.4-cudnn9-runtime
 
 ENV DEBIAN_FRONTEND=noninteractive \
@@ -22,9 +27,9 @@ RUN pip install --no-cache-dir 'kokoro>=0.9.0' 'fastapi>=0.115.0' 'uvicorn>=0.32
 
 RUN mkdir -p /cache/hf /audio
 
-COPY server.py /app/server.py
+COPY kokoro_server.py /app/kokoro_server.py
 WORKDIR /app
 
 EXPOSE 7860
 
-CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860"]
+CMD ["uvicorn", "kokoro_server:app", "--host", "0.0.0.0", "--port", "7860"]

engines/kokoro/compose.yml

@@ -1,18 +1,16 @@
-# Kokoro-82M TTS sidecar.
+# Kokoro-82M TTS stack on Lucy.
 #
-# Apache 2.0 code AND model weights — clean stack for share/publish.
-# Audiobook-quality narrator; F5-TTS stays around for voice-cloning.
+# Audiobook-quality narrator engine (Apache 2.0 code + weights —
+# clean stack vs F5-TTS's CC-BY-NC asterisk). Sibling to f5-tts;
+# both share /mnt/cache/appdata/f5-tts/audio so skald's audio
+# route serves outputs from either engine through the same path.
 #
-# Set in .env (or override):
-#   KOKORO_HOST_PORT=7794
-#   KOKORO_DATA=./data       # ${KOKORO_DATA}/hf-cache
-#   AUDIO_DIR=../f5-tts/data/audio  # shared output dir across engines
+# License: Apache 2.0 top to bottom. Right for share/publish.
 name: kokoro
 
 services:
   kokoro:
-    build: .
-    image: kokoro:0.5
+    image: lucy-registry:5000/kokoro:0.5
     container_name: kokoro
     restart: unless-stopped
     deploy:
@@ -23,10 +21,17 @@ services:
               count: all
               capabilities: [gpu]
     ports:
-      - "${KOKORO_HOST_PORT:-7794}:7860"
+      - "192.168.0.5:7794:7860"
+      - "127.0.0.1:7794:7860"
     volumes:
-      - ${KOKORO_DATA:-./data}/hf-cache:/cache/hf
-      - ${AUDIO_DIR:-./data/audio}:/audio
+      - /mnt/cache/appdata/kokoro/hf-cache:/cache/hf
+      # Shared with f5-tts so skald's /audio route covers both.
+      - /mnt/cache/appdata/f5-tts/audio:/audio
     environment:
       HF_HOME: /cache/hf
       HF_HUB_DISABLE_TELEMETRY: "1"
+    labels:
+      org.sulkta.domain: "sulkta"
+      org.sulkta.owner: "cobb"
+      org.sulkta.managed-by: "compose"
+      org.sulkta.role: "kokoro"

engines/kokoro/server.py

@@ -1,4 +1,4 @@
-"""Kokoro-82M FastAPI server, sibling to the f5-tts server.
+"""Kokoro-82M FastAPI server, sibling to f5_server.
 
 Same /synthesize contract as F5 so skald can route between engines
 just by which URL it points at. The semantic difference: Kokoro
@@ -234,7 +234,7 @@ def _startup() -> None:
 
 @app.get("/healthz")
 def healthz() -> dict:
-    # Shape matches the f5-tts server's so the same Rust HealthResponse
+    # Shape matches f5_server's so the same Rust HealthResponse
     # struct deserializes both: model/vocoder/loaded fields are
     # required by skald-core::narrate::HealthResponse.
     return {

engines/tortoise/Dockerfile

@@ -1,13 +1,17 @@
-# Tortoise-TTS. Apache 2.0 code + weights.
+# Sulkta build of Tortoise-TTS.
 #
-# ~26 built-in voices (no cloning): angie, daniel, deniro, emma,
-# freeman, geralt, halle, jlaw, lj, mol, myself, pat, pat2, rainbow,
-# snakes, tim_reynolds, tom, train_atkins, train_dotrice,
+# Voice roster (built-in, no cloning needed): angie, daniel, deniro,
+# emma, freeman, geralt, halle, jlaw, lj, mol, myself, pat, pat2,
+# rainbow, snakes, tim_reynolds, tom, train_atkins, train_dotrice,
 # train_dreams, train_grace, train_kennard, train_lescault,
-# train_mouse, weaver, william.
+# train_mouse, weaver, william. ~26 voices baked in.
 #
-# Slow: standard preset is ~10x slower than Kokoro; high_quality is
-# ~30x. Trade for quality.
+# License: Apache 2.0 (code) + Apache 2.0 (model weights). Clean
+# stack for share/publish.
+#
+# Speed: slow. Trade for quality. Standard preset is ~10x slower
+# than Kokoro; high_quality is ~30x slower. Worth it for the
+# audiobook-quality bar.
 
 FROM pytorch/pytorch:2.6.0-cuda12.4-cudnn9-runtime
 
@@ -33,9 +37,9 @@ RUN pip install --no-cache-dir \
 
 RUN mkdir -p /cache/hf /cache/tortoise-models /audio
 
-COPY server.py /app/server.py
+COPY tortoise_server.py /app/tortoise_server.py
 WORKDIR /app
 
 EXPOSE 7860
 
-CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860"]
+CMD ["uvicorn", "tortoise_server:app", "--host", "0.0.0.0", "--port", "7860"]

engines/tortoise/compose.yml

@@ -1,22 +1,19 @@
-# Tortoise-TTS sidecar. 25+ named voices, no cloning needed.
-# Apache 2.0 top to bottom.
+# Tortoise-TTS stack on Lucy. Audiobook-quality engine with 25+
+# named voices (no cloning). Apache 2.0 top to bottom.
 #
-# Slow: ~10x kokoro wall-clock at 'standard' preset. Worth it for the
-# quality bar; runs are batched.
+# Slow: ~10x kokoro wall clock at 'standard' preset. Worth it for
+# the quality bar. Cobb's call 2026-05-14: "use higgs (now tortoise)
+# and we will only let it use the full gpu for runs" — translated:
+# runs are batched, slow is acceptable.
 #
-# Co-resides with kokoro on an 8GB card (tortoise ~5GB + kokoro ~1GB).
-# OOM during a render: add a coordinator that pauses kokoro first.
-#
-# Set in .env (or override):
-#   TORTOISE_HOST_PORT=7795
-#   TORTOISE_DATA=./data     # ${TORTOISE_DATA}/{hf-cache,models}
-#   AUDIO_DIR=../f5-tts/data/audio  # shared output dir across engines
+# Co-resides with kokoro on the 2070 Super since tortoise is ~5GB
+# and kokoro is ~1GB (8GB total). If OOM hits during a render,
+# we'll add a coordination layer to pause kokoro first.
 name: tortoise
 
 services:
   tortoise:
-    build: .
-    image: tortoise:0.1
+    image: lucy-registry:5000/tortoise:0.1
     container_name: tortoise
     restart: unless-stopped
     deploy:
@@ -27,12 +24,20 @@ services:
               count: all
               capabilities: [gpu]
     ports:
-      - "${TORTOISE_HOST_PORT:-7795}:7860"
+      - "192.168.0.5:7795:7860"
+      - "127.0.0.1:7795:7860"
     volumes:
-      - ${TORTOISE_DATA:-./data}/hf-cache:/cache/hf
-      - ${TORTOISE_DATA:-./data}/models:/cache/tortoise-models
-      - ${AUDIO_DIR:-./data/audio}:/audio
+      - /mnt/cache/appdata/tortoise/hf-cache:/cache/hf
+      - /mnt/cache/appdata/tortoise/models:/cache/tortoise-models
+      # Shared audio dir with f5/kokoro so skald serves all engines'
+      # outputs through the same /audio route.
+      - /mnt/cache/appdata/f5-tts/audio:/audio
     environment:
       HF_HOME: /cache/hf
       HF_HUB_DISABLE_TELEMETRY: "1"
       TORTOISE_MODELS_DIR: /cache/tortoise-models
+    labels:
+      org.sulkta.domain: "sulkta"
+      org.sulkta.owner: "cobb"
+      org.sulkta.managed-by: "compose"
+      org.sulkta.role: "tortoise-tts"

engines/tortoise/exclusive-gpu.sh

@@ -0,0 +1,47 @@
+#!/bin/bash
+# Tortoise GPU exclusivity wrapper. The 2070 Super (8GB) can't host
+# F5 (~4.5GB) + Kokoro (~2.7GB) + Tortoise (~5GB peak) simultaneously,
+# so we stop the other two engines for the duration of a Tortoise run
+# and restart them after.
+#
+# Usage:
+#   exclusive-gpu.sh <command...>
+#
+# Example:
+#   exclusive-gpu.sh docker exec skald skald narrate --chapter <uuid>
+#
+# Exits with the wrapped command's status. Restarts the engines
+# regardless of success/failure (trap on EXIT).
+set -euo pipefail
+
+STOP_ENGINES=(f5-tts kokoro)
+
+cleanup() {
+    local rc=$?
+    echo "[exclusive-gpu] restarting engines"
+    for engine in "${STOP_ENGINES[@]}"; do
+        docker start "$engine" >/dev/null 2>&1 || \
+            echo "[exclusive-gpu] failed to restart $engine — investigate"
+    done
+    return "$rc"
+}
+trap cleanup EXIT
+
+echo "[exclusive-gpu] stopping engines: ${STOP_ENGINES[*]}"
+for engine in "${STOP_ENGINES[@]}"; do
+    docker stop "$engine" >/dev/null 2>&1 || true
+done
+
+# Restart Tortoise to clean up any cached GPU allocations from the
+# now-stopped engines (their CUDA contexts can linger briefly).
+docker restart tortoise >/dev/null
+echo "[exclusive-gpu] waiting for tortoise healthz..."
+for i in {1..30}; do
+    if curl -sf http://192.168.0.5:7795/healthz | grep -q '"loaded":true'; then
+        break
+    fi
+    sleep 2
+done
+
+echo "[exclusive-gpu] running: $*"
+"$@"

engines/tortoise/hacks.md

@@ -0,0 +1,71 @@
+# Tortoise engine — kludges branched off main
+
+This branch carries the engine-specific tweaks that don't generalise
+to F5 / Kokoro. Tortoise is the audiobook-quality engine but the
+trade-offs are real and need explicit handling — speed and GPU.
+
+## 1. GPU exclusivity
+
+**File:** `exclusive-gpu.sh`.
+
+The 2070 Super has 8GB. F5 (~4.5GB) + Kokoro (~2.7GB) + Tortoise
+(~5GB peak) sums to ~12GB — over budget. First Tortoise smoke
+caught it: `torch.OutOfMemoryError: ... 9.31 MiB is free`.
+
+Solution: stop the other two engines for the duration of a Tortoise
+run. The script wraps any command, stops `f5-tts` + `kokoro`,
+restarts `tortoise` to clean its CUDA context, waits for healthz,
+runs the wrapped command, then restarts the engines on EXIT trap
+(success or failure).
+
+```bash
+./exclusive-gpu.sh docker exec skald skald narrate --chapter <uuid>
+```
+
+Remove when: GPU upgrade (P40 24GB / 3090 24GB / etc) lets all three
+engines co-reside.
+
+## 2. Speed — slow, batch-only
+
+Tortoise at `standard` preset is **~74x slower than real-time** on
+the 2070 Super (smoke: 6.5s of audio took 478s wall clock). A 33-min
+Chapter 2 render would take ~8 hours. Tortoise is acceptable for
+overnight batched runs but NOT interactive rendering.
+
+Quality presets and their approx wall-clock for a 3000-word chapter:
+- `ultra_fast` — ~1h, noticeable quality drop
+- `fast` — ~2h
+- `standard` — ~6-8h, the recommended bar
+- `high_quality` — ~24h, marginally better than standard
+
+For most use, `standard` is right. Reserve `high_quality` for
+short prologues or named samples.
+
+## 3. Voice mapping format
+
+Tortoise's voice roster (`lj`, `freeman`, `daniel`, etc.) lives
+behind `source='tortoise_tts'` in the `voices` table. Character
+slug → Tortoise voice mapping is independent of the Kokoro mapping
+— a story can have BOTH a Kokoro and Tortoise mapping live in
+parallel, picked at render time via story.preferred_voice_id or
+the --voice flag.
+
+Tortoise voices may sometimes warble or stutter at chunk boundaries
+— the `tortoise.api.TextToSpeech.tts_with_preset` call is per-chunk
+and re-conditions the voice each time. Acceptable for v0.1; future
+work could feed `conditioning_latents` directly for tighter cohesion.
+
+## 4. No respelling overrides for Tortoise (yet)
+
+The `pronunciation_overrides` rows in the DB are seeded with
+lowercase-syllable respellings tuned for Kokoro's misaki tokenizer.
+Tortoise uses a different phonemizer (`g2p_en`) which handles many
+of those proper nouns better natively — but some still mangle.
+
+For now, narrate's substitution applies the same overrides regardless
+of engine, which means Tortoise sees `prip-yat` for "Pripyat" — same
+input, different phonemizer interprets differently. Usually OK but
+audit after each batch.
+
+Future: per-engine override sets, OR an `engine` column on
+pronunciation_overrides.

engines/tortoise/server.py

@@ -1,4 +1,4 @@
-"""Tortoise-TTS FastAPI server. Sibling to the kokoro server.
+"""Tortoise-TTS FastAPI server. Sibling to kokoro_server.
 
 Same /synthesize contract as the kokoro server so skald only has to
 route by voice.source. Differences:
@@ -23,6 +23,7 @@ import time
 import uuid
 from pathlib import Path
 
+import librosa
 import numpy as np
 import soundfile as sf
 import torch
@@ -62,28 +63,70 @@ def _get_tts() -> TextToSpeech:
     return _tts
 
 
+def _move_to_device(obj):
+    """Recursively .to(DEVICE) tensors inside the structure tortoise
+    returns from load_voice. voice_samples is a list of tensors;
+    conditioning_latents is a tuple of tensors. Anything else
+    passes through unchanged (e.g. None, ints)."""
+    if obj is None:
+        return obj
+    if isinstance(obj, torch.Tensor):
+        return obj.to(DEVICE)
+    if isinstance(obj, list):
+        return [_move_to_device(x) for x in obj]
+    if isinstance(obj, tuple):
+        return tuple(_move_to_device(x) for x in obj)
+    return obj
+
+
 def _get_voice(name: str) -> tuple:
     """Cache voice latents to avoid re-loading reference clips on
     every synthesis call. Tortoise's load_voice returns
-    (voice_samples, conditioning_latents)."""
+    (voice_samples, conditioning_latents) — but they're created on
+    CPU; we move them to DEVICE so the autoregressive model (on
+    CUDA) doesn't fail with cpu/cuda tensor-device mismatch."""
     if name not in _voice_cache:
-        _voice_cache[name] = load_voice(name)
+        samples, latents = load_voice(name)
+        _voice_cache[name] = (_move_to_device(samples), _move_to_device(latents))
     return _voice_cache[name]
 
 
-# ─── tag splitter (lifted from the kokoro server) ───────────────
+# ─── tag splitter (lifted from kokoro_server) ───────────────────
 
 
 class Node:
-    __slots__ = ("kind", "value", "voice")
+    __slots__ = ("kind", "value", "voice", "pitch", "rate")
 
-    def __init__(self, kind: str, value, voice: str | None = None):
+    def __init__(
+        self,
+        kind: str,
+        value,
+        voice: str | None = None,
+        pitch: float = 0.0,
+        rate: float = 1.0,
+    ):
+        # kind ∈ {"text", "silence"}; value is str for text, float
+        # seconds for silence. voice/pitch/rate are character-voicing
+        # modifiers from [voice:NAME pitch=N rate=R] tags. Default:
+        # request voice, 0 semitones, 1x rate.
         self.kind = kind
         self.value = value
         self.voice = voice
+        self.pitch = pitch
+        self.rate = rate
 
 
-_VOICE_OPEN_RE = re.compile(r"\[voice:([A-Za-z0-9_-]+)\]")
+# Voice open tag — name + optional pitch (semitones) + optional rate:
+#   [voice:dyatlov]               → voice swap only
+#   [voice:lj pitch=-3]           → same voice, 3 semitones lower
+#   [voice:lj pitch=2 rate=1.1]   → higher + slightly faster (fairy)
+#   [voice:lj pitch=-4 rate=0.9]  → lower + slower (troll)
+_VOICE_OPEN_RE = re.compile(
+    r"\[voice:([A-Za-z0-9_-]+)"
+    r"(?:\s+pitch=(-?[0-9]+(?:\.[0-9]+)?))?"
+    r"(?:\s+rate=([0-9]+(?:\.[0-9]+)?))?"
+    r"\]"
+)
 _VOICE_CLOSE = "[/voice]"
 _TAG_RE = re.compile(
     r"\[(pause:(?P<dur>[0-9]+(?:\.[0-9]+)?)(?P<unit>s|ms)?|breath|scene)\]",
@@ -102,7 +145,70 @@ def _parse_tag(match: re.Match) -> float:
     return dur / 1000.0 if unit == "ms" else dur
 
 
-def _expand_inline(text: str, voice: str | None) -> list[Node]:
+# Tortoise's autoregressive head loses coherence past ~20s of generated
+# audio per inference call. lj's pace is roughly 14 chars/s, so anything
+# past ~280 chars per call risks gibberish at the end. We split inside
+# _expand_inline at sentence boundaries to keep each tts_with_preset
+# call inside the model's reliable horizon.
+TORTOISE_MAX_CHUNK_CHARS = 220
+
+# Sentence boundary regex — splits on `.`/`?`/`!` followed by whitespace
+# and a capital letter (keeps "Mr. Smith" / "U.S." together) OR at any
+# newline.
+_SENTENCE_BOUNDARY = re.compile(r"(?<=[\.!?])\s+(?=[A-Z\"\(])|(?<=\n)\s*")
+
+
+def _chunk_for_tortoise(text: str, max_chars: int = TORTOISE_MAX_CHUNK_CHARS) -> list[str]:
+    """Split text into chunks <= max_chars at sentence boundaries.
+    If a single sentence exceeds max_chars (rare for prose), fall
+    back to splitting that sentence at commas or just hard-cutting.
+    """
+    sentences = [s.strip() for s in _SENTENCE_BOUNDARY.split(text) if s and s.strip()]
+    chunks: list[str] = []
+    current = ""
+    for sent in sentences:
+        # Long sentence: emit alone, but try sub-splitting at commas.
+        if len(sent) > max_chars:
+            if current:
+                chunks.append(current.strip())
+                current = ""
+            # Split on commas
+            parts = [p.strip() for p in sent.split(",") if p.strip()]
+            sub = ""
+            for p in parts:
+                add = (sub + ", " if sub else "") + p
+                if len(add) <= max_chars:
+                    sub = add
+                else:
+                    if sub:
+                        chunks.append(sub)
+                    # If even the part alone exceeds, hard-cut at max_chars
+                    while len(p) > max_chars:
+                        chunks.append(p[:max_chars])
+                        p = p[max_chars:]
+                    sub = p
+            if sub:
+                chunks.append(sub)
+            continue
+        # Sentence fits — accumulate.
+        candidate = (current + " " if current else "") + sent
+        if len(candidate) <= max_chars:
+            current = candidate
+        else:
+            if current:
+                chunks.append(current.strip())
+            current = sent
+    if current:
+        chunks.append(current.strip())
+    return chunks
+
+
+def _expand_inline(
+    text: str,
+    voice: str | None,
+    pitch: float = 0.0,
+    rate: float = 1.0,
+) -> list[Node]:
     out: list[Node] = []
     text = text.strip()
     if not text:
@@ -111,12 +217,12 @@ def _expand_inline(text: str, voice: str | None) -> list[Node]:
     for m in _TAG_RE.finditer(text):
         pre = text[cursor : m.start()].strip()
         if pre:
-            out.append(Node("text", pre, voice))
+            out.append(Node("text", pre, voice, pitch, rate))
         out.append(Node("silence", _parse_tag(m)))
         cursor = m.end()
     tail = text[cursor:].strip()
     if tail:
-        out.append(Node("text", tail, voice))
+        out.append(Node("text", tail, voice, pitch, rate))
     return out
 
 
@@ -130,12 +236,14 @@ def _split_paragraph_voices(para: str) -> list[Node]:
             break
         out.extend(_expand_inline(para[cursor : m.start()], None))
         voice = m.group(1)
+        pitch = float(m.group(2)) if m.group(2) else 0.0
+        rate = float(m.group(3)) if m.group(3) else 1.0
         body_start = m.end()
         close_idx = para.find(_VOICE_CLOSE, body_start)
         if close_idx < 0:
-            out.extend(_expand_inline(para[body_start:], voice))
+            out.extend(_expand_inline(para[body_start:], voice, pitch, rate))
             break
-        out.extend(_expand_inline(para[body_start:close_idx], voice))
+        out.extend(_expand_inline(para[body_start:close_idx], voice, pitch, rate))
         cursor = close_idx + len(_VOICE_CLOSE)
     return out
 
@@ -209,7 +317,7 @@ def _startup() -> None:
 
 @app.get("/healthz")
 def healthz() -> dict:
-    # Shape matches the f5-tts + kokoro servers so skald's HealthResponse
+    # Shape matches f5_server/kokoro_server so skald's HealthResponse
     # struct deserializes all three.
     return {
         "ok": True,
@@ -253,6 +361,7 @@ def synthesize(req: SynthesizeRequest) -> SynthesizeResponse:
     started = time.monotonic()
     pieces: list[np.ndarray] = []
     voices_used: set[str] = set()
+    tortoise_chunks_rendered = 0
     for node in nodes:
         if node.kind == "silence":
             pieces.append(_silence_samples(node.value))
@@ -264,18 +373,37 @@ def synthesize(req: SynthesizeRequest) -> SynthesizeResponse:
         except Exception as e:
             log.warning("voice %s failed to load (%s); falling back to default", seg_voice, e)
             samples, latents = _get_voice(voice)
-        # Tortoise's tts_with_preset returns a torch.Tensor on the
-        # configured device.
-        audio_tensor = tts.tts_with_preset(
-            text=node.value,
-            voice_samples=samples,
-            conditioning_latents=latents,
-            preset=preset,
-        )
-        if isinstance(audio_tensor, list):
-            audio_tensor = audio_tensor[0]
-        arr = audio_tensor.squeeze().cpu().numpy().astype(np.float32)
-        pieces.append(arr)
+        # Each text node may exceed Tortoise's reliable ~20s horizon —
+        # split at sentence boundaries before feeding the model.
+        sub_chunks = _chunk_for_tortoise(node.value)
+        for sub_idx, sub in enumerate(sub_chunks):
+            audio_tensor = tts.tts_with_preset(
+                text=sub,
+                voice_samples=samples,
+                conditioning_latents=latents,
+                preset=preset,
+            )
+            if isinstance(audio_tensor, list):
+                audio_tensor = audio_tensor[0]
+            arr = audio_tensor.squeeze().cpu().numpy().astype(np.float32)
+            # Per-character voice modulation via librosa. Apply
+            # pitch first (preserves duration), then rate (preserves
+            # pitch). Default pitch=0, rate=1.0 = no-op fast path.
+            if abs(node.pitch) > 1e-3:
+                arr = librosa.effects.pitch_shift(
+                    arr, sr=SAMPLE_RATE, n_steps=node.pitch
+                )
+            if abs(node.rate - 1.0) > 1e-3:
+                arr = librosa.effects.time_stretch(arr, rate=node.rate)
+            arr = arr.astype(np.float32)
+            pieces.append(arr)
+            tortoise_chunks_rendered += 1
+            log.info(
+                "chunk %d/%d done (%d chars, pitch=%+.1f rate=%.2f, %.1fs audio so far)",
+                sub_idx + 1, len(sub_chunks), len(sub),
+                node.pitch, node.rate,
+                sum(len(p) for p in pieces) / SAMPLE_RATE,
+            )
     elapsed_ms = int((time.monotonic() - started) * 1000)
 
     if not pieces:

entrypoint.sh

@@ -1,9 +1,14 @@
 #!/usr/bin/env bash
+# Skald container entrypoint.
+#
 # Boots the embedded postgres via the pgvector image's own
 # docker-entrypoint, waits for it to accept connections, then execs
-# `skald` in the foreground. Tini is PID 1; postgres becomes our
-# child. When the DB is extracted to its own container, this reduces
-# to `exec /usr/local/bin/skald "$@"`.
+# `skald` in the foreground. Tini is PID 1 (so it can reap zombies +
+# forward signals); we are PID 2; postgres becomes our child.
+#
+# This is explicitly "DB in the same container, for now" — when we
+# split the DB out (see project notes), the entrypoint reduces to
+# `exec /usr/local/bin/skald "$@"` and the pg startup goes away.
 
 set -eo pipefail
 

migrations/0008_chapter_rewrite.sql

@@ -1,15 +0,0 @@
--- The rewrite pass: an author re-authors existing chapter prose in
--- their own voice (canon preserved, prose reworked). body_md gets
--- overwritten with the rewritten version; body_md_original keeps
--- the pre-rewrite prose so the original is never lost. Populated
--- only on the FIRST rewrite of a chapter (if NULL) — subsequent
--- rewrites leave the original alone.
-ALTER TABLE chapters
-    ADD COLUMN IF NOT EXISTS body_md_original text;
-
--- Allow 'rewrite' as a generation_runs.kind.
-ALTER TABLE generation_runs
-    DROP CONSTRAINT generation_runs_kind_check;
-ALTER TABLE generation_runs
-    ADD CONSTRAINT generation_runs_kind_check
-    CHECK (kind = ANY (ARRAY['gen', 'cleanup', 'audit', 'summary', 'embed', 'narrate_prep', 'rewrite']));

migrations/0009_story_audiobook.sql

@@ -1,7 +0,0 @@
--- A story can have a single stitched audiobook — all its chapter
--- renders concatenated into one chaptered file (see the m4b built
--- from per-chapter narration_runs). audiobook_path is the path the
--- web server serves it from (e.g. /audio/The-Coast-Down.m4b); NULL
--- means no audiobook has been stitched yet.
-ALTER TABLE stories
-    ADD COLUMN IF NOT EXISTS audiobook_path text;

migrations/0010_prose_audit.sql

@@ -1,20 +0,0 @@
--- The prose-quality audit pass: a QC gate that reads a finished
--- story end to end and flags repetition, template tics, self-
--- restatement and continuity drift before it goes to narration.
--- Repetition a silent reader skims is glaring once narrated aloud.
-
--- Allow 'prose_audit' as a generation_runs.kind.
-ALTER TABLE generation_runs DROP CONSTRAINT generation_runs_kind_check;
-ALTER TABLE generation_runs ADD CONSTRAINT generation_runs_kind_check
-    CHECK (kind = ANY (ARRAY[
-        'gen', 'cleanup', 'audit', 'summary', 'embed',
-        'narrate_prep', 'rewrite', 'prose_audit'
-    ]));
-
--- 'repetition' is a first-class audit finding area.
-ALTER TABLE audit_findings DROP CONSTRAINT audit_findings_area_check;
-ALTER TABLE audit_findings ADD CONSTRAINT audit_findings_area_check
-    CHECK (area = ANY (ARRAY[
-        'character', 'continuity', 'tone', 'fact',
-        'timeline', 'repetition', 'other'
-    ]));

migrations/0011_dedup_pass.sql

@@ -1,9 +0,0 @@
--- The dedup pass: the fix half of the audit loop. Takes a chapter
--- plus the story's prose-audit findings and rephrases only the
--- flagged repetitions, leaving everything else verbatim.
-ALTER TABLE generation_runs DROP CONSTRAINT generation_runs_kind_check;
-ALTER TABLE generation_runs ADD CONSTRAINT generation_runs_kind_check
-    CHECK (kind = ANY (ARRAY[
-        'gen', 'cleanup', 'audit', 'summary', 'embed',
-        'narrate_prep', 'rewrite', 'prose_audit', 'dedup'
-    ]));

skald-core/src/config.rs

@@ -8,23 +8,24 @@ use serde::{Deserialize, Serialize};
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct ForgeConfig {
-    /// Base URL of the clawdforge HTTP service. The calling binary
-    /// resolves this from `CLAWDFORGE_URL`.
+    /// Base URL of the clawdforge HTTP service. Defaults to
+    /// `http://clawdforge.sulkta.lan:8800` in production; override
+    /// for tests via env.
     pub base_url: String,
 
     /// App-level bearer token. Resolved by the binary from
     /// `CLAWDFORGE_TOKEN`; should never be logged or `Display`ed.
     pub app_token: String,
 
-    /// Model alias passed to clawdforge → `claude -p --model`.
-    /// Defaults to opus.
+    /// Model alias passed to clawdforge → `claude -p --model`. Skald
+    /// is opinionated: always opus max effort. Default reflects that.
     pub model: String,
 }
 
 impl Default for ForgeConfig {
     fn default() -> Self {
         Self {
-            base_url: "http://localhost:8800".into(),
+            base_url: "http://clawdforge.sulkta.lan:8800".into(),
             app_token: String::new(),
             model: "opus".into(),
         }

skald-core/src/forge.rs

@@ -7,7 +7,7 @@
 //!
 //! 1. **gen** — produces a new chapter draft from an assembled
 //!    context blob (parent prose + bible + characters + similarity-
-//!    matched passages, all from the database). Opus, high effort.
+//!    matched passages, all from the database). Opus, max effort.
 //!
 //! 2. **cleanup** — polishes the draft for prose quality, voice
 //!    consistency, dialogue rhythm, pacing dead spots. Same Opus,
@@ -43,11 +43,8 @@ use crate::config::ForgeConfig;
 pub struct Forge {
     client: Client,
     /// The model alias we pass to clawdforge. Skald is opinionated:
-    /// always opus. Story-writing passes (gen/cleanup/narrate_prep/
-    /// rewrite) run at HIGH effort; only the audit pass runs at MAX —
-    /// audit genuinely needs the frontier reasoning, prose-craft does
-    /// not, and the $200/mo `claude -p` cap makes max-everywhere
-    /// unaffordable. `clawdforge` resolves the alias to the CLI flag.
+    /// always opus max effort. (See `project_story_writer_container.md`.)
+    /// `clawdforge` resolves the alias to the actual claude CLI flag.
     model: String,
 }
 
@@ -77,18 +74,6 @@ pub enum PassKind {
     /// prose; output should be byte-identical except for the
     /// tag insertions.
     NarratePrep,
-    /// Re-author existing chapter prose in an author's voice. Canon
-    /// (names, dates, events, places, facts) is preserved exactly;
-    /// the prose itself is rewritten. Not editing — re-authoring.
-    Rewrite,
-    /// Whole-story prose-quality audit — reads a finished story end
-    /// to end and flags repetition, template tics, self-restatement
-    /// and continuity drift. The QC gate before narration.
-    ProseAudit,
-    /// Surgical dedup — takes a chapter plus the story's audit
-    /// findings and rephrases only the flagged repetitions, leaving
-    /// everything else verbatim. The fix half of the audit loop.
-    Dedup,
 }
 
 impl PassKind {
@@ -99,9 +84,6 @@ impl PassKind {
             Self::Audit => "audit",
             Self::Summary => "summary",
             Self::NarratePrep => "narrate_prep",
-            Self::Rewrite => "rewrite",
-            Self::ProseAudit => "prose_audit",
-            Self::Dedup => "dedup",
         }
     }
 }
@@ -111,12 +93,10 @@ impl Forge {
         let client = ClientBuilder::default()
             .base_url(&cfg.base_url)
             .token(&cfg.app_token)
-            // Generation passes at --effort max run 10–20 min wall clock;
-            // the max-effort whole-book prose audit can run past an hour.
-            // This is the client-side ceiling — set high enough to cover
-            // the slowest pass; each pass still sends its own server-side
-            // timeout_secs. Default 120s would strand any real pass.
-            .timeout(Duration::from_secs(7200))
+            // Generation passes at --effort max can run 10–20 min wall
+            // clock. clawdforge's server-side cap is 1800s — match it.
+            // Default 120s would strand any prose-craft pass.
+            .timeout(Duration::from_secs(1800))
             .user_agent(concat!("skald/", env!("CARGO_PKG_VERSION")))
             .build()?;
         Ok(Self {
@@ -151,7 +131,7 @@ impl Forge {
             model: Some(self.model.clone()),
             system: Some(system),
             system_mode: Some(mode),
-            effort: Some(Effort::High),
+            effort: Some(Effort::Max),
             timeout_secs: Some(1800),
             ..Default::default()
         };
@@ -176,7 +156,7 @@ impl Forge {
             model: Some(self.model.clone()),
             system: Some(system),
             system_mode: Some(mode),
-            effort: Some(Effort::High),
+            effort: Some(Effort::Max),
             timeout_secs: Some(1800),
             ..Default::default()
         };
@@ -205,75 +185,6 @@ impl Forge {
         })
     }
 
-    /// Whole-story prose-quality audit. Receives every chapter of a
-    /// finished story concatenated in order and returns findings JSON
-    /// — repetition, template tics, self-restatement, continuity
-    /// drift. This is the QC gate before a story goes to narration,
-    /// where repetition a silent reader skims becomes glaring.
-    ///
-    /// HIGH effort: finding repeated passages and motif reuse is
-    /// comparison work, not deep reasoning — high-effort Opus does it
-    /// well in minutes. A max-effort pass over a whole-book input
-    /// (quarter-million chars) runs an impractical hour-plus.
-    pub async fn prose_audit(&self, full_story: &str) -> anyhow::Result<PassOutput> {
-        let prompt = format!(
-            "Audit the complete story below for repetition, template tics, \
-             self-restatement and continuity errors. Read every chapter. \
-             Return JSON only, matching the schema in the system prompt.\n\n\
-             {full_story}"
-        );
-        let body = RunRequest {
-            prompt,
-            model: Some(self.model.clone()),
-            system: Some(SYSTEM_PROSE_AUDIT.to_string()),
-            effort: Some(Effort::High),
-            // High effort over a whole book lands in minutes; 3600s is
-            // a generous ceiling that won't bite.
-            timeout_secs: Some(3600),
-            ..Default::default()
-        };
-        let r = self.client.run(body).await?;
-        let duration_ms = r.duration_ms;
-        Ok(PassOutput { kind: PassKind::ProseAudit, result: r, duration_ms })
-    }
-
-    /// Surgical dedup of one chapter — the fix half of the audit
-    /// loop. Receives the chapter's prose plus the whole story's
-    /// audit findings, and rephrases ONLY the flagged repetitions
-    /// that occur in this chapter; everything the findings do not
-    /// flag stays verbatim. Author REQUIRED — the fresh phrasing
-    /// lands in the author's voice (SystemMode::Replace). High
-    /// effort: it is prose-craft, same posture as rewrite.
-    pub async fn dedup(
-        &self,
-        prose: &str,
-        findings: &str,
-        author: &AuthorWithRevision,
-    ) -> anyhow::Result<PassOutput> {
-        let scaffold = author
-            .revision
-            .system_template
-            .as_deref()
-            .unwrap_or(DEFAULT_AUTHOR_SCAFFOLD);
-        let system = scaffold
-            .replace("{{display_name}}", &author.author.display_name)
-            .replace("{{pass_directive}}", DEDUP_DIRECTIVE)
-            .replace("{{soul}}", &author.revision.soul);
-        let user_prompt = dedup_user_prompt(prose, findings);
-        let body = RunRequest {
-            prompt: user_prompt,
-            model: Some(self.model.clone()),
-            system: Some(system),
-            system_mode: Some(SystemMode::Replace),
-            effort: Some(Effort::High),
-            timeout_secs: Some(1800),
-            ..Default::default()
-        };
-        let r = self.client.run(body).await?;
-        let duration_ms = r.duration_ms;
-        Ok(PassOutput { kind: PassKind::Dedup, result: r, duration_ms })
-    }
-
     /// Annotate prose with narration control tags. The model
     /// receives the full chapter prose and returns the SAME prose
     /// with `[pause:Xs]`, `[breath]`, `[scene]` markers inserted
@@ -281,12 +192,10 @@ impl Forge {
     /// Orson Black places beats differently than another author
     /// would. Replace-mode if author is set; Append otherwise.
     ///
-    /// `characters` is the story's character roster. When non-empty,
+    /// `characters` is the story's character roster. When provided,
     /// the system prompt instructs the model to wrap dialogue in
-    /// `[voice:<slug>]"..."[/voice]` for multi-voice rendering; the
+    /// `[voice:<slug>]"..."[/voice]` for multi-voice rendering. The
     /// slug is mapped to a Kokoro voice id by skald's narrate path.
-    /// An EMPTY roster selects single-voice mode — the prompt then
-    /// forbids `[voice:...]` tags entirely (one narrator, no cast).
     ///
     /// Hard rule the system prompt enforces: do not change a word
     /// of prose. Tags are additive only.
@@ -296,12 +205,6 @@ impl Forge {
         author: Option<&AuthorWithRevision>,
         characters: &[CharacterSpeaker],
     ) -> anyhow::Result<PassOutput> {
-        // An empty character roster means single-voice narration —
-        // the whole chapter reads in one voice. In that mode the
-        // prompt must NOT invite `[voice:...]` tags, or the model
-        // invents speaker slugs from names in the prose that the
-        // narrate path then has to detect and neutralize.
-        let single_voice = characters.is_empty();
         let user_prompt = narrate_prep_user_prompt(prose, characters);
         let (system, mode) = match author {
             Some(a) => {
@@ -310,34 +213,22 @@ impl Forge {
                     .system_template
                     .as_deref()
                     .unwrap_or(DEFAULT_AUTHOR_SCAFFOLD);
-                let directive = if single_voice {
-                    NARRATE_PREP_DIRECTIVE_SINGLE
-                } else {
-                    NARRATE_PREP_DIRECTIVE
-                };
                 let composed = scaffold
                     .replace("{{display_name}}", &a.author.display_name)
-                    .replace("{{pass_directive}}", directive)
+                    .replace("{{pass_directive}}", NARRATE_PREP_DIRECTIVE)
                     .replace("{{soul}}", &a.revision.soul);
                 (composed, SystemMode::Replace)
             }
-            None => {
-                let house = if single_voice {
-                    HOUSE_NARRATE_PREP_SYSTEM_SINGLE
-                } else {
-                    HOUSE_NARRATE_PREP_SYSTEM
-                };
-                (house.to_string(), SystemMode::Append)
-            }
+            None => (HOUSE_NARRATE_PREP_SYSTEM.to_string(), SystemMode::Append),
         };
         let body = RunRequest {
             prompt: user_prompt,
             model: Some(self.model.clone()),
             system: Some(system),
             system_mode: Some(mode),
-            // Tag placement IS a craft choice; high effort is
-            // plenty for beat sense. Same posture as gen/cleanup.
-            effort: Some(Effort::High),
+            // Tag placement IS a craft choice; max effort buys
+            // better beat sense. Same posture as gen/cleanup.
+            effort: Some(Effort::Max),
             timeout_secs: Some(1800),
             ..Default::default()
         };
@@ -346,46 +237,6 @@ impl Forge {
         Ok(PassOutput { kind: PassKind::NarratePrep, result: r, duration_ms })
     }
 
-    /// Re-author existing chapter prose in the author's voice. The
-    /// model receives prose written by another hand and rewrites it
-    /// entirely in its own style — sentence rhythm, word choice,
-    /// paragraph shape all become the author's. Canon is preserved
-    /// exactly: names, dates, events, places, technical facts, and
-    /// the sequence of what happens do not change.
-    ///
-    /// Author REQUIRED — a rewrite without an author has no target
-    /// voice. SystemMode::Replace; the model BECOMES the author.
-    /// High effort: re-authoring is heavy prose-craft, but it's
-    /// still craft, not reasoning — max is reserved for audit.
-    pub async fn rewrite(
-        &self,
-        prose: &str,
-        author: &AuthorWithRevision,
-    ) -> anyhow::Result<PassOutput> {
-        let scaffold = author
-            .revision
-            .system_template
-            .as_deref()
-            .unwrap_or(DEFAULT_AUTHOR_SCAFFOLD);
-        let system = scaffold
-            .replace("{{display_name}}", &author.author.display_name)
-            .replace("{{pass_directive}}", REWRITE_DIRECTIVE)
-            .replace("{{soul}}", &author.revision.soul);
-        let user_prompt = rewrite_user_prompt(prose);
-        let body = RunRequest {
-            prompt: user_prompt,
-            model: Some(self.model.clone()),
-            system: Some(system),
-            system_mode: Some(SystemMode::Replace),
-            effort: Some(Effort::High),
-            timeout_secs: Some(1800),
-            ..Default::default()
-        };
-        let r = self.client.run(body).await?;
-        let duration_ms = r.duration_ms;
-        Ok(PassOutput { kind: PassKind::Rewrite, result: r, duration_ms })
-    }
-
     /// Summarize one chapter to ~250 words. The summary feeds into
     /// the continuation context for older chapters so the token
     /// budget stays sane on long series (book 12 doesn't carry book 1
@@ -484,38 +335,20 @@ Hard rules:
 {{soul}}
 "#;
 
-const GEN_DIRECTIVE: &str = "This is a GENERATION pass. Write the next chapter from scratch. Honor canon. Begin with a chapter heading on the first line.\n\nGuard against repetition. If a recurring image, motif or descriptive beat appeared in an earlier chapter, render it in fresh words here — never reuse the same sentence shape or verb sequence for it. Do not stack consecutive sentences on one template (\"He thought… He thought…\", \"She felt… She felt…\"). Vary how you land beats. This will be read aloud by a single narrator, so phrasing a silent reader skims is glaring to a listener.";
+const GEN_DIRECTIVE: &str = "This is a GENERATION pass. Write the next chapter from scratch. Honor canon. Begin with a chapter heading on the first line.";
 
-const CLEANUP_DIRECTIVE: &str = "This is a CLEANUP pass. The user prompt contains a draft you wrote. Polish for prose quality — tighten dialogue, fix pacing dead spots, hold your voice steady. Do NOT add new plot, do NOT retcon canon.\n\nHunt repetition hard — this prose will be narrated aloud, where repetition is far more glaring than on the page. If a recurring image or motif appears more than once, re-phrase every occurrence so no two share a sentence shape or verb sequence. Do not stack more than two consecutive sentences on the same opening stem (\"He thought… He thought…\"). Echo a line of dialogue back verbatim at most once in the whole chapter. If a sentence merely restates something already said, cut it. Return ONLY the polished chapter.";
+const CLEANUP_DIRECTIVE: &str = "This is a CLEANUP pass. The user prompt contains a draft you wrote. Polish for prose quality — tighten dialogue, fix pacing dead spots, hold your voice steady. Do NOT add new plot, do NOT retcon canon. Return ONLY the polished chapter.";
 
 const HOUSE_GEN_SYSTEM: &str = "You are a long-form fiction author writing the next chapter of a series. Honor the canon (characters, setting, established facts) exactly. Return only the chapter prose, starting with a heading line. No preamble.";
 
 const HOUSE_CLEANUP_SYSTEM: &str = "You are a copy editor polishing a draft chapter. Tighten dialogue, fix pacing, keep voice consistent. Do not add new plot. Return only the polished chapter.";
 
-const SYSTEM_PROSE_AUDIT: &str = "You are a ruthless prose-quality auditor for long-form fiction destined for single-voice audiobook narration. You receive a complete story — every chapter, in order — and you read all of it. You return STRUCTURED JSON ONLY: no commentary, no preamble. You hunt four things:\n\n1. REPETITION across the whole book: a recurring image, motif or descriptive beat rendered in near-identical wording more than once; a sentence (or near-identical sentence) reused in different chapters; a line of dialogue echoed verbatim.\n2. TEMPLATE TICS: the same sentence template stacked or over-reused — e.g. clusters of consecutive sentences all opening 'He thought:' / 'She felt:' — anywhere it becomes a noticeable pattern.\n3. SELF-RESTATEMENT: a sentence or paragraph that says again, slightly reworded, something the text already said.\n4. CONTINUITY ERRORS: a fact, name, age, date or detail that contradicts an earlier one.\n\nThis prose will be read ALOUD by one narrator, so repetition a silent reader skims is glaring to a listener — be exhaustive, do not let near-duplicates pass. Return EXACTLY this shape: { \"findings\": [ { \"severity\": \"info\"|\"warn\"|\"crit\", \"area\": \"repetition\"|\"continuity\"|\"other\", \"body\": \"...\" } ] }. In each finding's body, quote the offending text exactly and name the chapter number(s). Use 'crit' for anything a listener will plainly hear as a mistake, 'warn' for noticeable repetition, 'info' for minor. If the story is clean, return { \"findings\": [] }.";
-
 const SYSTEM_AUDIT: &str = "You are a canon auditor for long-form fiction. You compare a parent story and a new chapter against the bible. You flag continuity drift, character voice shift, retconned facts, dropped threads, timeline contradictions. You return STRUCTURED JSON ONLY — no commentary, no preamble. The exact shape: { \"findings\": [ { \"severity\": \"info\"|\"warn\"|\"crit\", \"area\": \"character\"|\"continuity\"|\"tone\"|\"fact\"|\"timeline\"|\"other\", \"body\": \"...\" } ] }. If no findings, return { \"findings\": [] }.";
 
 const NARRATE_PREP_DIRECTIVE: &str = "This is a NARRATION-ANNOTATION pass. You receive your own prose and prepare it for an audiobook reading. Three kinds of inserts are allowed:\n\n1. BEAT MARKERS (additive, not prose): `[breath]` (~400ms), `[pause:1.2s]` (explicit silence in seconds, e.g. 0.5s, 1.2s, 2s), `[scene]` (~1500ms scene break). Place where the prose's rhythm asks for them — after a hard one-line beat, before a turn in dialogue, on a paragraph that lands with weight.\n\n2. SPEAKER VOICE TAGS (multi-voice dialogue): wrap dialogue lines in `[voice:<slug>]\"...\"[/voice]` based on who is speaking. The roster of available speaker slugs is given in the user prompt. The dialogue itself stays verbatim — only the wrapper is added. If a line of dialogue is not clearly attributable to a roster speaker, leave it unwrapped (the narrator voice will read it). Quoted thoughts (italicized interior monologue) stay unwrapped — only spoken aloud dialogue gets a voice tag.\n\n3. NARRATOR STUMBLES (humanizing prose-level inserts): a real narrator occasionally stumbles on a hard word, catches themselves, repeats. You may add these *sparingly* where the prose's pacing makes them feel right. Patterns: em-dash repetition (`Prip— Pripyat`), self-correction (`she — no, the wife — had been told`), hesitation (`the dose, the dose was`). USE SPARINGLY. Maybe 1-3 per chapter. Pick proper nouns, technical terms, or moments where the narrator might genuinely catch herself. Avoid stumbling on emotional climaxes — those should land clean.\n\nApart from stumbles, do NOT change a word of the original prose. Return the prose with beat markers, voice tags, and stumbles inline. No preamble. No commentary about your choices.";
 
-/// Single-voice variant of [`NARRATE_PREP_DIRECTIVE`]. Used when the
-/// chapter narrates in one voice (no speaker roster). The multi-voice
-/// directive's section 2 is dropped entirely AND a hard prohibition
-/// is added — without it the model invents `[voice:<slug>]` tags from
-/// character names in the prose, which the narrate path then has to
-/// detect and neutralize.
-const NARRATE_PREP_DIRECTIVE_SINGLE: &str = "This is a NARRATION-ANNOTATION pass. You receive your own prose and prepare it for a SINGLE-narrator audiobook reading — the whole chapter, dialogue included, is read aloud in ONE voice. Two kinds of inserts are allowed:\n\n1. BEAT MARKERS (additive, not prose): `[breath]` (~400ms), `[pause:1.2s]` (explicit silence in seconds, e.g. 0.5s, 1.2s, 2s), `[scene]` (~1500ms scene break). Place where the prose's rhythm asks for them — after a hard one-line beat, before a turn in dialogue, on a paragraph that lands with weight.\n\n2. NARRATOR STUMBLES (humanizing prose-level inserts): a real narrator occasionally stumbles on a hard word, catches themselves, repeats. You may add these *sparingly* where the prose's pacing makes them feel right. Patterns: em-dash repetition (`Prip— Pripyat`), self-correction (`she — no, the wife — had been told`), hesitation (`the dose, the dose was`). USE SPARINGLY. Maybe 1-3 per chapter. Pick proper nouns, technical terms, or moments where the narrator might genuinely catch herself. Avoid stumbling on emotional climaxes — those should land clean.\n\nDo NOT add `[voice:...]` speaker tags of any kind — there is one narrator, not a cast. Apart from stumbles, do NOT change a word of the original prose. Return the prose with beat markers and stumbles inline. No preamble. No commentary about your choices.";
-
 const HOUSE_NARRATE_PREP_SYSTEM: &str = "You are a senior audiobook director annotating prose for narration. You insert (a) beat markers — `[breath]`, `[pause:Xs]`, `[scene]` — where a skilled narrator would breathe or pause, (b) speaker voice tags `[voice:<slug>]\"...\"[/voice]` wrapping dialogue based on who is speaking (roster supplied in user prompt; leave unattributed dialogue unwrapped), and (c) occasional humanizing narrator stumbles using em-dash repetition or self-correction (sparingly — maybe 1-3 per chapter, on proper nouns or hard words). Apart from those stumbles you do NOT change a word of the prose. Return the prose verbatim plus beat markers, voice tags, and (rare) stumbles inline. No preamble, no commentary.";
 
-/// Single-voice variant of [`HOUSE_NARRATE_PREP_SYSTEM`] — no speaker
-/// voice tags, one narrator throughout.
-const HOUSE_NARRATE_PREP_SYSTEM_SINGLE: &str = "You are a senior audiobook director annotating prose for a SINGLE-narrator reading. You insert (a) beat markers — `[breath]`, `[pause:Xs]`, `[scene]` — where a skilled narrator would breathe or pause, and (b) occasional humanizing narrator stumbles using em-dash repetition or self-correction (sparingly — maybe 1-3 per chapter, on proper nouns or hard words). Do NOT add `[voice:...]` speaker tags — the whole chapter is one voice. Apart from those stumbles you do NOT change a word of the prose. Return the prose verbatim plus beat markers and (rare) stumbles inline. No preamble, no commentary.";
-
-const DEDUP_DIRECTIVE: &str = "This is a DEDUP pass. The user prompt contains ONE chapter of a story you wrote, plus a list of audit findings — repeated phrases, motifs, similes, sentence templates and continuity errors found across the whole book. Your job: return this chapter with every flagged repetition that occurs IN IT rephrased fresh, and everything else byte-identical.\n\nHARD RULES:\n- For any motif, simile, phrase, image or structural tic the findings flag as recurring: if it appears in THIS chapter, render this chapter's occurrence in fresh, distinctive wording. Never reuse the flagged original phrasing. The other chapters' occurrences are being revised separately — do NOT try to coordinate with them; just make yours distinct from the flagged original.\n- Fix any continuity error the findings flag that touches this chapter (a wrong age, number, name, date) — use the correct value the findings identify.\n- Change NOTHING the findings do not flag. Every sentence not implicated by a finding stays EXACTLY as written, word for word. This is not a rewrite, not a polish, not an edit for taste — it is a surgical dedup. When in doubt, leave it.\n- Canon is absolute: names, dates, events, the order they happen, every fact — unchanged. The chapter stays the same length and shape.\n- Return ONLY the chapter prose. No heading unless the source had one. No preamble, no commentary, no list of what you changed.\n\n";
-
-const REWRITE_DIRECTIVE: &str = "This is a REWRITE pass. The user prompt contains a chapter of prose written by another hand. Re-author it entirely in YOUR voice — every sentence reworked in your style: your sentence rhythm, your word choice, your paragraph shape, your way of landing a beat. This is not editing or polishing. It is re-authoring. The reader should not be able to tell another writer ever touched it.\n\nHARD CONSTRAINTS — canon is non-negotiable:\n- Every character name, every date, every place name stays exactly as written.\n- Every event, and the ORDER events happen in, stays exactly as written.\n- Every technical or historical fact stays exactly as written.\n- Do not add new scenes, characters, or events. Do not cut any scene or beat. Same story, same shape — your telling.\n\nReturn ONLY the rewritten chapter prose. Begin with the chapter heading line (`## Chapter N — title`) exactly as in the source. No preamble, no commentary about the rewrite.";
-
 // ─── User-prompt builders ───────────────────────────────────────
 
 fn gen_user_prompt(
@@ -562,38 +395,6 @@ pub struct CharacterSpeaker {
     pub hint: Option<String>,
 }
 
-fn dedup_user_prompt(prose: &str, findings: &str) -> String {
-    let mut out = String::with_capacity(prose.len() + findings.len() + 512);
-    out.push_str("# Audit findings for this story\n\n");
-    out.push_str(
-        "These repetitions and errors were found across the whole book. \
-         Fix only the ones that occur in the chapter below.\n\n",
-    );
-    out.push_str(findings);
-    out.push_str("\n\n# Chapter to dedup\n\n");
-    out.push_str(prose);
-    out.push_str(
-        "\n\n# Task\n\nReturn the chapter above with every flagged repetition \
-         that appears in it rephrased fresh, and any flagged continuity error \
-         touching it corrected. Leave every unflagged sentence verbatim. \
-         Return only the chapter prose.\n",
-    );
-    out
-}
-
-fn rewrite_user_prompt(prose: &str) -> String {
-    let mut out = String::with_capacity(prose.len() + 256);
-    out.push_str("# Chapter to re-author\n\n");
-    out.push_str(prose);
-    out.push_str(
-        "\n\n# Task\n\nRe-author the chapter above entirely in your voice. \
-         Preserve all canon — names, dates, places, events, the order they \
-         happen, every technical fact. Change only the prose. Return only \
-         the rewritten chapter, starting with its `## Chapter N` heading.\n",
-    );
-    out
-}
-
 fn narrate_prep_user_prompt(prose: &str, characters: &[CharacterSpeaker]) -> String {
     let mut out = String::with_capacity(prose.len() + 512);
 
@@ -623,22 +424,12 @@ fn narrate_prep_user_prompt(prose: &str, characters: &[CharacterSpeaker]) -> Str
 
     out.push_str("# Prose to annotate\n\n");
     out.push_str(prose);
-    if characters.is_empty() {
-        out.push_str(
-            "\n\n# Task\n\nReturn the prose above with `[breath]`, `[pause:Xs]`, \
-             `[scene]` beat markers inserted appropriately. Do NOT add any \
-             `[voice:...]` tags — this is a single-voice reading. Do not \
-             change any word. Do not skip any sentence. Return only the \
-             annotated prose.\n",
-        );
-    } else {
-        out.push_str(
-            "\n\n# Task\n\nReturn the prose above with `[breath]`, `[pause:Xs]`, \
-             `[scene]` markers and `[voice:<slug>]\"...\"[/voice]` dialogue wrappers \
-             inserted appropriately. Do not change any word. Do not skip any \
-             sentence. Return only the annotated prose.\n",
-        );
-    }
+    out.push_str(
+        "\n\n# Task\n\nReturn the prose above with `[breath]`, `[pause:Xs]`, \
+         `[scene]` markers and `[voice:<slug>]\"...\"[/voice]` dialogue wrappers \
+         inserted appropriately. Do not change any word. Do not skip any \
+         sentence. Return only the annotated prose.\n",
+    );
     out
 }
 
@@ -666,10 +457,6 @@ fn build_audit_request(model: &str, parent: &str, sequel: &str, bible: &str) ->
         prompt,
         model: Some(model.to_string()),
         system: Some(SYSTEM_AUDIT.to_string()),
-        // Audit is the one pass that keeps MAX effort — catching
-        // canon drift, timeline gaps and retcons is reasoning work
-        // worth the frontier spend; prose-craft passes run at high.
-        effort: Some(Effort::Max),
         timeout_secs: Some(600),
         ..Default::default()
     }

skald-core/src/narrate.rs

@@ -27,7 +27,7 @@ use uuid::Uuid;
 
 #[derive(Debug, Clone)]
 pub struct F5Config {
-    /// e.g. http://localhost:7792
+    /// e.g. http://192.168.0.5:7792
     pub base_url: String,
     /// Inference subprocess timeout. Long-form chapters (3000 words)
     /// take 60-180s on an 8GB GPU; cap at 1800s to match clawdforge.

skald/src/audit.rs

@@ -1,144 +0,0 @@
-//! `skald audit` — whole-story prose-quality audit. Reads every
-//! chapter of a story end to end and flags repetition, template
-//! tics, self-restatement and continuity drift. The QC gate before
-//! a story goes to narration, where repetition a silent reader
-//! skims becomes glaring once read aloud.
-//!
-//! Findings land in the `audit_findings` table (area 'repetition' /
-//! 'continuity' / 'other') and are printed to stdout. v1 flags
-//! only — acting on the findings (a fix pass) is a separate step.
-
-use std::time::Instant;
-
-use anyhow::{Context, bail};
-use chrono::Utc;
-use skald_core::config::ForgeConfig;
-use skald_core::db;
-use skald_core::forge::{AuditFinding, AuditResponse, Forge, PassKind, PassOutput};
-use uuid::Uuid;
-
-pub async fn run(database_url: &str, story_id: Uuid) -> anyhow::Result<()> {
-    let cfg = load_forge_config()?;
-    tracing::info!(base_url = %cfg.base_url, model = %cfg.model, "forge configured");
-
-    let pool = db::connect_and_migrate(database_url).await?;
-    let forge = Forge::new(&cfg)?;
-
-    let title: String = sqlx::query_scalar("SELECT title FROM stories WHERE id = $1")
-        .bind(story_id)
-        .fetch_optional(&pool)
-        .await?
-        .with_context(|| format!("story {story_id} not found"))?;
-
-    let chapters: Vec<(i32, Option<String>, String)> = sqlx::query_as(
-        "SELECT n, title, body_md FROM chapters WHERE story_id = $1 ORDER BY n",
-    )
-    .bind(story_id)
-    .fetch_all(&pool)
-    .await?;
-    if chapters.is_empty() {
-        bail!("story {story_id} has no chapters to audit");
-    }
-
-    // Concatenate the whole story in chapter order — the audit needs
-    // to see across chapter boundaries to catch cross-chapter reuse.
-    let mut full = String::new();
-    for (n, ct, body) in &chapters {
-        full.push_str(&format!(
-            "## Chapter {n} — {}\n\n",
-            ct.as_deref().unwrap_or("")
-        ));
-        full.push_str(body);
-        full.push_str("\n\n");
-    }
-    tracing::info!(
-        story = %title,
-        chapters = chapters.len(),
-        chars = full.len(),
-        "prose audit starting",
-    );
-
-    let run_id: Uuid = sqlx::query_scalar(
-        "INSERT INTO generation_runs (story_id, kind, status) VALUES ($1, $2, 'running') RETURNING id",
-    )
-    .bind(story_id)
-    .bind(PassKind::ProseAudit.as_str())
-    .fetch_one(&pool)
-    .await?;
-
-    let started = Instant::now();
-    let out_res = forge.prose_audit(&full).await;
-    let elapsed = started.elapsed();
-
-    let out: PassOutput = match out_res {
-        Ok(o) => o,
-        Err(e) => {
-            sqlx::query(
-                "UPDATE generation_runs SET status='failed', error=$1, ended_at=$2 WHERE id=$3",
-            )
-            .bind(format!("{e:#}"))
-            .bind(Utc::now())
-            .bind(run_id)
-            .execute(&pool)
-            .await?;
-            return Err(e);
-        }
-    };
-
-    let findings = parse_findings(&out);
-    for f in &findings {
-        sqlx::query(
-            "INSERT INTO audit_findings (story_id, run_id, severity, area, body)
-             VALUES ($1, $2, $3, $4, $5)",
-        )
-        .bind(story_id)
-        .bind(run_id)
-        .bind(&f.severity)
-        .bind(&f.area)
-        .bind(&f.body)
-        .execute(&pool)
-        .await?;
-    }
-    sqlx::query("UPDATE generation_runs SET status='succeeded', ended_at=$1 WHERE id=$2")
-        .bind(Utc::now())
-        .bind(run_id)
-        .execute(&pool)
-        .await?;
-
-    let crit = findings.iter().filter(|f| f.severity == "crit").count();
-    let warn = findings.iter().filter(|f| f.severity == "warn").count();
-    let info = findings.iter().filter(|f| f.severity == "info").count();
-    println!(
-        "prose audit: \"{title}\" — {} finding(s): {crit} crit, {warn} warn, {info} info ({:.1}s)",
-        findings.len(),
-        elapsed.as_secs_f32(),
-    );
-    for f in &findings {
-        println!("\n[{} · {}]\n{}", f.severity.to_uppercase(), f.area, f.body);
-    }
-    Ok(())
-}
-
-fn parse_findings(out: &PassOutput) -> Vec<AuditFinding> {
-    if let Ok(typed) = out.result.as_json::<AuditResponse>() {
-        return typed.findings;
-    }
-    if let Some(s) = out.result.as_text() {
-        if let Ok(typed) = serde_json::from_str::<AuditResponse>(s) {
-            return typed.findings;
-        }
-    }
-    tracing::warn!("prose audit output did not parse as AuditResponse — no findings recorded");
-    Vec::new()
-}
-
-fn load_forge_config() -> anyhow::Result<ForgeConfig> {
-    let base_url = std::env::var("CLAWDFORGE_URL").context("CLAWDFORGE_URL not set")?;
-    let app_token = std::env::var("CLAWDFORGE_TOKEN").context("CLAWDFORGE_TOKEN not set")?;
-    let model = std::env::var("SKALD_MODEL").unwrap_or_else(|_| "opus".into());
-    Ok(ForgeConfig {
-        base_url,
-        app_token,
-        model,
-    })
-}

skald/src/dedup.rs

@@ -1,271 +0,0 @@
-//! `skald dedup` — the fix half of the audit loop. Reads a story's
-//! most recent prose-audit findings and walks the chapters, handing
-//! each chapter + the findings to the author with instructions to
-//! rephrase ONLY the flagged repetitions and leave everything else
-//! verbatim. Overwrites body_md and clears body_md_tts so the
-//! chapter gets re-prepped before narration.
-//!
-//! Run `skald audit` first — dedup needs findings to act on.
-
-use std::time::Instant;
-
-use anyhow::{Context, bail};
-use chrono::Utc;
-use skald_core::authors::{self, AuthorWithRevision};
-use skald_core::config::ForgeConfig;
-use skald_core::db;
-use skald_core::forge::{Forge, PassKind, PassOutput};
-use sqlx::PgPool;
-use uuid::Uuid;
-
-pub async fn run(
-    database_url: &str,
-    story_id: Uuid,
-    author_slug: Option<&str>,
-    chapter_filter: Option<i32>,
-) -> anyhow::Result<()> {
-    let cfg = load_forge_config()?;
-    tracing::info!(base_url = %cfg.base_url, model = %cfg.model, "forge configured");
-
-    let pool = db::connect_and_migrate(database_url).await?;
-    let forge = Forge::new(&cfg)?;
-
-    let story: Option<(String, Option<Uuid>)> =
-        sqlx::query_as("SELECT title, author_id FROM stories WHERE id = $1")
-            .bind(story_id)
-            .fetch_optional(&pool)
-            .await?;
-    let (title, story_author_id) =
-        story.with_context(|| format!("story {story_id} not found"))?;
-
-    let author = resolve_author(&pool, story_author_id, author_slug)
-        .await?
-        .ok_or_else(|| {
-            anyhow::anyhow!(
-                "dedup needs an author for the rephrasing — pass --author <slug> \
-                 or bind one to the story"
-            )
-        })?;
-
-    // Findings from the most recent succeeded prose-audit run.
-    let findings = load_latest_findings(&pool, story_id).await?;
-    if findings.is_empty() {
-        bail!(
-            "no prose-audit findings for story {story_id} — run `skald audit \
-             --story {story_id}` first"
-        );
-    }
-    let findings_block = render_findings(&findings);
-    tracing::info!(
-        story = %title,
-        author = %author.author.slug,
-        finding_count = findings.len(),
-        "dedup starting",
-    );
-
-    let mut chapters: Vec<(Uuid, i32)> = sqlx::query_as(
-        "SELECT id, n FROM chapters WHERE story_id = $1 ORDER BY n",
-    )
-    .bind(story_id)
-    .fetch_all(&pool)
-    .await?;
-    if chapters.is_empty() {
-        bail!("story {story_id} has no chapters");
-    }
-    // --chapter narrows the run to one chapter — used to retry a
-    // chapter the length guard skipped on an earlier run.
-    if let Some(target) = chapter_filter {
-        chapters.retain(|(_, n)| *n == target);
-        if chapters.is_empty() {
-            bail!("chapter {target} not found in story {story_id}");
-        }
-    }
-
-    let mut skipped = 0usize;
-    for (chapter_id, n) in &chapters {
-        let body_md: String =
-            sqlx::query_scalar("SELECT body_md FROM chapters WHERE id = $1")
-                .bind(chapter_id)
-                .fetch_one(&pool)
-                .await?;
-
-        let run_id: Uuid = sqlx::query_scalar(
-            "INSERT INTO generation_runs (story_id, kind, status) VALUES ($1, $2, 'running') RETURNING id",
-        )
-        .bind(story_id)
-        .bind(PassKind::Dedup.as_str())
-        .fetch_one(&pool)
-        .await?;
-
-        let started = Instant::now();
-        let out_res = forge.dedup(&body_md, &findings_block, &author).await;
-        let elapsed = started.elapsed();
-
-        let out: PassOutput = match out_res {
-            Ok(o) => o,
-            Err(e) => {
-                sqlx::query(
-                    "UPDATE generation_runs SET status='failed', error=$1, ended_at=$2 WHERE id=$3",
-                )
-                .bind(format!("{e:#}"))
-                .bind(Utc::now())
-                .bind(run_id)
-                .execute(&pool)
-                .await?;
-                return Err(e).with_context(|| format!("dedup failed on chapter {n}"));
-            }
-        };
-
-        let deduped = pass_text(&out)?;
-
-        // Sanity guard: a surgical dedup nudges a chapter's length by
-        // a little. An output wildly off the input means the model
-        // duplicated or ballooned the chapter — reject it, leave the
-        // chapter untouched, move on. A re-run with --chapter retries
-        // just the skipped one.
-        let before = body_md.len();
-        let after = deduped.len();
-        if after > before * 3 / 2 || after < before * 3 / 5 {
-            sqlx::query(
-                "UPDATE generation_runs SET status='failed', error=$1, ended_at=$2 WHERE id=$3",
-            )
-            .bind(format!(
-                "rejected: dedup output {after}c is wildly off input {before}c \
-                 — likely a duplicated or ballooned output"
-            ))
-            .bind(Utc::now())
-            .bind(run_id)
-            .execute(&pool)
-            .await?;
-            skipped += 1;
-            println!(
-                "SKIPPED chapter {n}: dedup returned {after}c from {before}c \
-                 — chapter left untouched (retry with --chapter {n})"
-            );
-            continue;
-        }
-
-        // Overwrite body_md and clear body_md_tts — the chapter must be
-        // re-prepped before it is narrated again. body_md_original is
-        // left untouched (it belongs to the rewrite pass).
-        sqlx::query("UPDATE chapters SET body_md = $1, body_md_tts = NULL WHERE id = $2")
-            .bind(&deduped)
-            .bind(chapter_id)
-            .execute(&pool)
-            .await?;
-        sqlx::query("UPDATE generation_runs SET status='succeeded', ended_at=$1 WHERE id=$2")
-            .bind(Utc::now())
-            .bind(run_id)
-            .execute(&pool)
-            .await?;
-
-        println!(
-            "deduped chapter {n} ({before}c -> {after}c) in {:.1}s",
-            elapsed.as_secs_f32(),
-        );
-    }
-
-    if skipped > 0 {
-        println!(
-            "dedup complete: \"{title}\" — {} chapter(s) processed, {skipped} SKIPPED \
-             (retry each with --chapter)",
-            chapters.len(),
-        );
-    } else {
-        println!(
-            "dedup complete: \"{title}\" — {} chapter(s) deduped against {} finding(s)",
-            chapters.len(),
-            findings.len(),
-        );
-    }
-    Ok(())
-}
-
-#[derive(Debug, Clone)]
-struct Finding {
-    severity: String,
-    area: String,
-    body: String,
-}
-
-async fn load_latest_findings(pool: &PgPool, story_id: Uuid) -> anyhow::Result<Vec<Finding>> {
-    let rows: Vec<(String, String, String)> = sqlx::query_as(
-        "SELECT severity, area, body FROM audit_findings
-         WHERE story_id = $1
-           AND run_id = (
-             SELECT id FROM generation_runs
-             WHERE story_id = $1 AND kind = 'prose_audit' AND status = 'succeeded'
-             ORDER BY started_at DESC LIMIT 1
-           )
-         ORDER BY
-           CASE severity WHEN 'crit' THEN 0 WHEN 'warn' THEN 1 ELSE 2 END,
-           area",
-    )
-    .bind(story_id)
-    .fetch_all(pool)
-    .await?;
-    Ok(rows
-        .into_iter()
-        .map(|(severity, area, body)| Finding { severity, area, body })
-        .collect())
-}
-
-fn render_findings(findings: &[Finding]) -> String {
-    let mut out = String::new();
-    for f in findings {
-        out.push_str(&format!(
-            "[{} · {}]\n{}\n\n",
-            f.severity.to_uppercase(),
-            f.area,
-            f.body,
-        ));
-    }
-    out
-}
-
-async fn resolve_author(
-    pool: &PgPool,
-    story_author_id: Option<Uuid>,
-    flag_slug: Option<&str>,
-) -> anyhow::Result<Option<AuthorWithRevision>> {
-    if let Some(slug) = flag_slug {
-        return authors::get_with_current_revision(pool, slug)
-            .await?
-            .map(Some)
-            .with_context(|| format!("author '{slug}' not found"));
-    }
-    if let Some(aid) = story_author_id {
-        let row: Option<(String,)> = sqlx::query_as("SELECT slug FROM authors WHERE id = $1")
-            .bind(aid)
-            .fetch_optional(pool)
-            .await?;
-        if let Some((slug,)) = row {
-            return Ok(authors::get_with_current_revision(pool, &slug).await?);
-        }
-    }
-    Ok(None)
-}
-
-fn pass_text(out: &PassOutput) -> anyhow::Result<String> {
-    let text = out
-        .result
-        .as_text()
-        .map(|s| s.to_string())
-        .or_else(|| out.result.result.as_str().map(|s| s.to_string()))
-        .unwrap_or_else(|| out.result.result.to_string());
-    if text.trim().is_empty() {
-        bail!("dedup pass returned empty");
-    }
-    Ok(text)
-}
-
-fn load_forge_config() -> anyhow::Result<ForgeConfig> {
-    let base_url = std::env::var("CLAWDFORGE_URL").context("CLAWDFORGE_URL not set")?;
-    let app_token = std::env::var("CLAWDFORGE_TOKEN").context("CLAWDFORGE_TOKEN not set")?;
-    let model = std::env::var("SKALD_MODEL").unwrap_or_else(|_| "opus".into());
-    Ok(ForgeConfig {
-        base_url,
-        app_token,
-        model,
-    })
-}

skald/src/main.rs

@@ -4,14 +4,11 @@
 //!   skald serve              — boot the http server (v0.1 = /health + migrations)
 //!   skald import-markdown    — ingest a story markdown file into the DB
 
-mod audit;
 mod authors_seed;
 mod continue_story;
-mod dedup;
 mod import;
 mod narrate;
 mod narrate_prep;
-mod rewrite;
 mod serve;
 mod show_context;
 mod summarize;
@@ -30,8 +27,8 @@ use uuid::Uuid;
     about = "Long-form story-writer. Database is the source of truth; the writer is the tooling."
 )]
 struct Cli {
-    /// Postgres connection URL. Read from `DATABASE_URL` if unset.
-    #[arg(long, env = "DATABASE_URL")]
+    /// Postgres connection URL. Defaults to `postgresql://skald:skald@localhost:5432/skald`.
+    #[arg(long, env = "DATABASE_URL", default_value = "postgresql://skald:skald@localhost:5432/skald")]
     database_url: String,
 
     #[command(subcommand)]
@@ -158,51 +155,6 @@ enum Cmd {
         /// errors out to avoid clobbering a hand-tuned version.
         #[arg(long)]
         overwrite: bool,
-        /// Single-voice mode: skip the character speaker roster so
-        /// no [voice:X] dialogue tags are inserted. Use when the
-        /// whole chapter narrates in one voice.
-        #[arg(long)]
-        single_voice: bool,
-    },
-    /// Re-author one chapter's prose in an author's voice. Canon
-    /// preserved, prose reworked. Overwrites body_md (stashing the
-    /// original in body_md_original) and clears body_md_tts.
-    Rewrite {
-        /// Chapter UUID to re-author.
-        #[arg(long)]
-        chapter: Uuid,
-        /// Author slug to rewrite as. Falls back to the story's
-        /// bound author if omitted.
-        #[arg(long)]
-        author: Option<String>,
-    },
-    /// Whole-story prose-quality audit. Reads every chapter end to
-    /// end and flags repetition, template tics, self-restatement
-    /// and continuity drift. The QC gate before narration. Findings
-    /// land in audit_findings and print to stdout. Requires
-    /// CLAWDFORGE_URL + CLAWDFORGE_TOKEN.
-    Audit {
-        /// Story to audit.
-        #[arg(long)]
-        story: Uuid,
-    },
-    /// Dedup a story against its most recent prose-audit findings.
-    /// Walks every chapter, rephrasing only the flagged repetitions
-    /// and fixing flagged continuity errors — everything else stays
-    /// verbatim. Overwrites body_md and clears body_md_tts. Run
-    /// `skald audit` first.
-    Dedup {
-        /// Story to dedup.
-        #[arg(long)]
-        story: Uuid,
-        /// Author slug for the rephrasing. Falls back to the story's
-        /// bound author if omitted.
-        #[arg(long)]
-        author: Option<String>,
-        /// Restrict the run to a single chapter number — used to
-        /// retry a chapter the length guard skipped.
-        #[arg(long)]
-        chapter: Option<i32>,
     },
 }
 
@@ -278,23 +230,8 @@ async fn run() -> anyhow::Result<()> {
             chapter,
             author,
             overwrite,
-            single_voice,
         } => {
-            narrate_prep::run(
-                &cli.database_url,
-                chapter,
-                author.as_deref(),
-                overwrite,
-                single_voice,
-            )
-            .await
-        }
-        Cmd::Rewrite { chapter, author } => {
-            rewrite::run(&cli.database_url, chapter, author.as_deref()).await
-        }
-        Cmd::Audit { story } => audit::run(&cli.database_url, story).await,
-        Cmd::Dedup { story, author, chapter } => {
-            dedup::run(&cli.database_url, story, author.as_deref(), chapter).await
+            narrate_prep::run(&cli.database_url, chapter, author.as_deref(), overwrite).await
         }
     }
 }

skald/src/narrate.rs

@@ -65,16 +65,12 @@ pub async fn run(
     let run_id = Uuid::new_v4();
     let output_filename = format!("{}-{}-{}.wav", chapter.story_id, chapter.n, run_id);
 
-    // Engine + version threaded from the voice row's source. Three
-    // engines on Lucy currently:
-    //   kokoro_*    → kokoro 82M
-    //   tortoise_*  → tortoise-tts
-    //   anything else (lj_speech etc.) → f5-tts
-    // Future: a dedicated voices.engine column to make this explicit.
+    // Engine + version threaded from the voice row's source/license
+    // pair. lj_speech-style PD voices live behind f5-tts; kokoro_*
+    // voices live behind kokoro. Future: a dedicated voices.engine
+    // column to make this explicit.
     let (engine, engine_version) = if voice.source.starts_with("kokoro") {
         ("kokoro-82m", "0.9")
-    } else if voice.source.starts_with("tortoise") {
-        ("tortoise-tts", "3.0")
     } else {
         ("f5-tts", "1.1.20")
     };
@@ -95,20 +91,14 @@ pub async fn run(
     // the Kokoro server only ever sees real voice ids. Only kicks
     // in for kokoro-routed renders; F5 voice-tag handling isn't
     // implemented and any tags pass through unchanged.
-    // Two pre-processing passes (kokoro + tortoise — engines that
-    // parse [voice:X] dialogue tags). Order matters:
-    //   1. Speaker voice substitution rewrites [voice:slug] → the
-    //      engine's named voice id. Must run BEFORE pronunciation
-    //      overrides so we don't try to respell character slugs.
-    //      Tortoise: characters with no tortoise-voice mapping
-    //      gracefully fall back to the narrator voice server-side.
+    // Two pre-processing passes (kokoro only). Order matters:
+    //   1. Speaker voice substitution rewrites [voice:slug] → [voice:kokoro_id].
+    //      This must run BEFORE pronunciation overrides so we don't
+    //      accidentally try to respell character slugs.
     //   2. Pronunciation overrides word-substitute proper nouns
-    //      (Pripyat, Dyatlov, etc.) with English-readable
-    //      respellings. The respellings are kokoro/misaki-tuned but
-    //      pass through tortoise's g2p_en well enough to apply.
-    let routes_to_engine_with_voice_tags =
-        voice.source.starts_with("kokoro") || voice.source.starts_with("tortoise");
-    let gen_text = if routes_to_engine_with_voice_tags {
+    //      (Pripyat, Dyatlov, etc.) with English-readable respellings
+    //      so Kokoro's small phonemizer doesn't mangle them.
+    let gen_text = if voice.source.starts_with("kokoro") {
         let voiced = substitute_speaker_voices(
             &pool,
             chapter.story_id,
@@ -161,12 +151,6 @@ pub async fn run(
     .execute(&pool)
     .await?;
 
-    // This chapter now has a fresh canonical render. Prior render
-    // WAVs are dead weight — every re-render otherwise leaves its
-    // predecessor on disk forever. Reclaim it. Best-effort: a
-    // cleanup failure must never fail an otherwise-good render.
-    cleanup_superseded_renders(&pool, chapter_id, run_row_id).await;
-
     println!(
         "narrated chapter {} of story {}: {} ({:.2}s audio, {:.1}s wall clock)",
         chapter.n,
@@ -389,84 +373,15 @@ async fn apply_pronunciation_overrides(
     Ok(out)
 }
 
-/// Delete the WAV files of prior renders of this chapter and clear
-/// their `output_path`. The newest succeeded render is the canonical
-/// one; older renders are superseded the moment a new one lands, and
-/// without this every re-render would leave a stale ~80MB file on
-/// disk forever.
-///
-/// The `narration_runs` rows themselves are KEPT — engine, voice,
-/// timing and status stay as render history. Only `output_path` is
-/// nulled, so no row ever points at a file that no longer exists.
-///
-/// Best-effort throughout: this runs *after* the current render has
-/// already been recorded as succeeded, so any failure here (a query
-/// error, a permission problem on the audio dir) is logged and
-/// swallowed — it must never turn a good render into a failed one.
-async fn cleanup_superseded_renders(pool: &PgPool, chapter_id: Uuid, current_run: Uuid) {
-    // output_path is only ever set on the success UPDATE, so
-    // "output_path IS NOT NULL AND id != current" is exactly the set
-    // of prior completed renders.
-    let prior: Vec<(Uuid, String)> = match sqlx::query_as(
-        "SELECT id, output_path FROM narration_runs
-         WHERE chapter_id = $1 AND id <> $2 AND output_path IS NOT NULL",
-    )
-    .bind(chapter_id)
-    .bind(current_run)
-    .fetch_all(pool)
-    .await
-    {
-        Ok(rows) => rows,
-        Err(e) => {
-            tracing::warn!(error = %e, "superseded-render cleanup: query failed, skipping");
-            return;
-        }
-    };
-
-    for (run_id, output_path) in prior {
-        // output_path is the HTTP-facing path "/audio/<file>"; the
-        // `/audio` bind mount means that is also the on-disk path
-        // inside this container.
-        match std::fs::remove_file(&output_path) {
-            Ok(()) => {
-                tracing::info!(run_id = %run_id, path = %output_path, "removed superseded render");
-            }
-            Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
-                // File already gone — still clear the dangling row.
-            }
-            Err(e) => {
-                // Could not delete — leave output_path intact rather
-                // than pointing the row at nothing.
-                tracing::warn!(
-                    run_id = %run_id, path = %output_path, error = %e,
-                    "superseded-render cleanup: could not delete file, leaving row intact",
-                );
-                continue;
-            }
-        }
-        if let Err(e) = sqlx::query("UPDATE narration_runs SET output_path = NULL WHERE id = $1")
-            .bind(run_id)
-            .execute(pool)
-            .await
-        {
-            tracing::warn!(run_id = %run_id, error = %e, "superseded-render cleanup: could not null output_path");
-        }
+/// Pick the engine base URL for a given voice.source. Voices whose
+/// source starts with "kokoro" route to KOKORO_URL; everything else
+/// routes to F5_TTS_URL. Each env var has a LAN-default for Lucy.
+fn engine_url_for(source: &str) -> anyhow::Result<String> {
+    if source.starts_with("kokoro") {
+        Ok(std::env::var("KOKORO_URL")
+            .unwrap_or_else(|_| "http://192.168.0.5:7794".into()))
+    } else {
+        Ok(std::env::var("F5_TTS_URL")
+            .unwrap_or_else(|_| "http://192.168.0.5:7792".into()))
     }
 }
-
-/// Pick the engine base URL for a given voice.source.
-///   kokoro_*    → KOKORO_URL
-///   tortoise_*  → TORTOISE_URL
-///   anything else (lj_speech etc.) → F5_TTS_URL
-fn engine_url_for(source: &str) -> anyhow::Result<String> {
-    let (env_var, engine) = if source.starts_with("kokoro") {
-        ("KOKORO_URL", "kokoro")
-    } else if source.starts_with("tortoise") {
-        ("TORTOISE_URL", "tortoise")
-    } else {
-        ("F5_TTS_URL", "f5-tts")
-    };
-    std::env::var(env_var).map_err(|_| {
-        anyhow::anyhow!("{env_var} not set — point at the {engine} sidecar")
-    })
-}

skald/src/narrate_prep.rs

@@ -24,7 +24,6 @@ pub async fn run(
     chapter_id: Uuid,
     author_slug: Option<&str>,
     overwrite: bool,
-    single_voice: bool,
 ) -> anyhow::Result<()> {
     let cfg = load_forge_config()?;
     tracing::info!(base_url = %cfg.base_url, model = %cfg.model, "forge configured");
@@ -61,20 +60,10 @@ pub async fn run(
     .fetch_one(&pool)
     .await?;
 
-    // Single-voice mode skips the speaker roster entirely — the
-    // narrate_prep pass then inserts only [breath]/[pause]/[scene]
-    // beats, no [voice:X] dialogue tags. Right when the whole
-    // chapter narrates in one voice.
-    let characters = if single_voice {
-        tracing::info!("single-voice mode — skipping speaker roster");
-        Vec::new()
-    } else {
-        let c = load_speakers(&pool, chapter.story_id).await?;
-        if !c.is_empty() {
-            tracing::info!(speaker_count = c.len(), "speaker roster loaded");
-        }
-        c
-    };
+    let characters = load_speakers(&pool, chapter.story_id).await?;
+    if !characters.is_empty() {
+        tracing::info!(speaker_count = characters.len(), "speaker roster loaded");
+    }
 
     let started = Instant::now();
     let out_res = forge

skald/src/rewrite.rs

@@ -1,252 +0,0 @@
-//! `skald rewrite` — re-author one chapter's prose in an author's
-//! voice. Canon preserved, prose reworked. Overwrites chapters.body_md
-//! with the rewritten version; the pre-rewrite prose is stashed in
-//! chapters.body_md_original on the first rewrite (if NULL) so the
-//! original is never lost.
-//!
-//! Author resolution: --author flag wins, else the chapter's
-//! story.author_id. A rewrite with no author errors — there's no
-//! target voice.
-
-use std::time::Instant;
-
-use anyhow::{Context, bail};
-use chrono::Utc;
-use skald_core::authors::{self, AuthorWithRevision};
-use skald_core::config::ForgeConfig;
-use skald_core::db;
-use skald_core::forge::{Forge, PassKind, PassOutput};
-use sqlx::PgPool;
-use uuid::Uuid;
-
-pub async fn run(
-    database_url: &str,
-    chapter_id: Uuid,
-    author_slug: Option<&str>,
-) -> anyhow::Result<()> {
-    let cfg = load_forge_config()?;
-    tracing::info!(base_url = %cfg.base_url, model = %cfg.model, "forge configured");
-
-    let pool = db::connect_and_migrate(database_url).await?;
-    let forge = Forge::new(&cfg)?;
-
-    let chapter = load_chapter(&pool, chapter_id).await?;
-    let author = resolve_author(&pool, &chapter, author_slug)
-        .await?
-        .ok_or_else(|| {
-            anyhow::anyhow!(
-                "rewrite needs an author — pass --author <slug> or bind one to the story"
-            )
-        })?;
-    tracing::info!(
-        slug = %author.author.slug,
-        revision_n = author.revision.n,
-        chapter_n = chapter.n,
-        word_count_in = word_count(&chapter.body_md),
-        "re-authoring chapter",
-    );
-
-    let run_id: Uuid = sqlx::query_scalar(
-        "INSERT INTO generation_runs (story_id, kind, status) VALUES ($1, $2, 'running') RETURNING id",
-    )
-    .bind(chapter.story_id)
-    .bind(PassKind::Rewrite.as_str())
-    .fetch_one(&pool)
-    .await?;
-
-    let started = Instant::now();
-    let out_res = forge.rewrite(&chapter.body_md, &author).await;
-    let elapsed = started.elapsed();
-
-    let out: PassOutput = match out_res {
-        Ok(o) => o,
-        Err(e) => {
-            sqlx::query(
-                "UPDATE generation_runs SET status='failed', error=$1, ended_at=$2 WHERE id=$3",
-            )
-            .bind(format!("{e:#}"))
-            .bind(Utc::now())
-            .bind(run_id)
-            .execute(&pool)
-            .await?;
-            return Err(e);
-        }
-    };
-
-    let rewritten = pass_text(&out)?;
-    let (_n, title, body) = parse_chapter(&rewritten);
-
-    // Stash the original on first rewrite, then overwrite body_md.
-    // body_md_tts is cleared — it was annotated against the OLD
-    // prose and must be regenerated by a fresh prepare-narration.
-    sqlx::query(
-        "UPDATE chapters
-         SET body_md_original = COALESCE(body_md_original, body_md),
-             body_md = $1,
-             title = COALESCE($2, title),
-             body_md_tts = NULL,
-             word_count = $3,
-             generated_at = now()
-         WHERE id = $4",
-    )
-    .bind(&body)
-    .bind(title.as_deref())
-    .bind(word_count(&body))
-    .bind(chapter_id)
-    .execute(&pool)
-    .await?;
-
-    // Replace passages with the rewritten paragraphs.
-    sqlx::query("DELETE FROM passages WHERE chapter_id = $1")
-        .bind(chapter_id)
-        .execute(&pool)
-        .await?;
-    for (i, para) in body.split("\n\n").enumerate() {
-        let p = para.trim();
-        if p.is_empty() || p == "---" {
-            continue;
-        }
-        sqlx::query("INSERT INTO passages (chapter_id, paragraph_n, body) VALUES ($1, $2, $3)")
-            .bind(chapter_id)
-            .bind(i as i32 + 1)
-            .bind(p)
-            .execute(&pool)
-            .await?;
-    }
-    sqlx::query(
-        "UPDATE stories SET word_count_actual = (SELECT COALESCE(SUM(word_count), 0) FROM chapters WHERE story_id = $1) WHERE id = $1",
-    )
-    .bind(chapter.story_id)
-    .execute(&pool)
-    .await?;
-
-    sqlx::query("UPDATE generation_runs SET status='succeeded', ended_at=$1 WHERE id=$2")
-        .bind(Utc::now())
-        .bind(run_id)
-        .execute(&pool)
-        .await?;
-
-    println!(
-        "rewrote chapter {} of story {} as {} ({} → {} words) in {:.1}s",
-        chapter.n,
-        chapter.story_id,
-        author.author.slug,
-        word_count(&chapter.body_md),
-        word_count(&body),
-        elapsed.as_secs_f32(),
-    );
-    Ok(())
-}
-
-#[derive(Debug, Clone)]
-struct ChapterRow {
-    story_id: Uuid,
-    n: i32,
-    body_md: String,
-    story_author_id: Option<Uuid>,
-}
-
-async fn load_chapter(pool: &PgPool, id: Uuid) -> anyhow::Result<ChapterRow> {
-    let row: Option<(Uuid, i32, String, Option<Uuid>)> = sqlx::query_as(
-        "SELECT c.story_id, c.n, c.body_md, s.author_id
-         FROM chapters c JOIN stories s ON s.id = c.story_id
-         WHERE c.id = $1",
-    )
-    .bind(id)
-    .fetch_optional(pool)
-    .await?;
-    let (story_id, n, body_md, story_author_id) =
-        row.with_context(|| format!("chapter {id} not found"))?;
-    Ok(ChapterRow {
-        story_id,
-        n,
-        body_md,
-        story_author_id,
-    })
-}
-
-async fn resolve_author(
-    pool: &PgPool,
-    chapter: &ChapterRow,
-    flag_slug: Option<&str>,
-) -> anyhow::Result<Option<AuthorWithRevision>> {
-    if let Some(slug) = flag_slug {
-        return authors::get_with_current_revision(pool, slug)
-            .await?
-            .map(Some)
-            .with_context(|| format!("author '{slug}' not found"));
-    }
-    if let Some(aid) = chapter.story_author_id {
-        let row: Option<(String,)> = sqlx::query_as("SELECT slug FROM authors WHERE id = $1")
-            .bind(aid)
-            .fetch_optional(pool)
-            .await?;
-        if let Some((slug,)) = row {
-            return Ok(authors::get_with_current_revision(pool, &slug).await?);
-        }
-    }
-    Ok(None)
-}
-
-fn pass_text(out: &PassOutput) -> anyhow::Result<String> {
-    let text = out
-        .result
-        .as_text()
-        .map(|s| s.to_string())
-        .or_else(|| out.result.result.as_str().map(|s| s.to_string()))
-        .unwrap_or_else(|| out.result.result.to_string());
-    if text.trim().is_empty() {
-        bail!("rewrite pass returned empty");
-    }
-    Ok(text)
-}
-
-/// Parse (n, title, body) out of the rewritten chapter. Tolerant of
-/// a missing heading — if the first line isn't a heading we keep the
-/// whole text as body and return n=0 (caller keeps the existing n).
-fn parse_chapter(text: &str) -> (i32, Option<String>, String) {
-    let trimmed = text.trim_start();
-    let first = trimmed.lines().next().unwrap_or("").trim();
-    if let Some(heading) = first.strip_prefix('#') {
-        let heading = heading.trim_start_matches('#').trim();
-        let n = heading
-            .to_lowercase()
-            .find("chapter")
-            .and_then(|idx| {
-                heading[idx + 7..]
-                    .trim_start()
-                    .split([' ', '—', '-', ':', ','])
-                    .next()
-                    .and_then(|w| w.parse::<i32>().ok())
-            })
-            .unwrap_or(0);
-        let title = heading
-            .split_once(" — ")
-            .or_else(|| heading.split_once(" - "))
-            .map(|(_, t)| t.trim().to_string())
-            .filter(|t| !t.is_empty());
-        let body = trimmed
-            .lines()
-            .skip(1)
-            .collect::<Vec<_>>()
-            .join("\n")
-            .trim_start()
-            .to_string();
-        let body = if body.is_empty() { text.trim().to_string() } else { body };
-        return (n, title, body);
-    }
-    (0, None, text.trim().to_string())
-}
-
-fn word_count(s: &str) -> i32 {
-    s.split_whitespace().count() as i32
-}
-
-fn load_forge_config() -> anyhow::Result<ForgeConfig> {
-    let base_url = std::env::var("CLAWDFORGE_URL")
-        .context("CLAWDFORGE_URL not set")?;
-    let app_token = std::env::var("CLAWDFORGE_TOKEN")
-        .context("CLAWDFORGE_TOKEN not set")?;
-    let model = std::env::var("SKALD_MODEL").unwrap_or_else(|_| "opus".into());
-    Ok(ForgeConfig { base_url, app_token, model })
-}

skald/src/web.rs

@@ -45,7 +45,6 @@ pub fn router(state: WebState) -> Router {
             "/stories/{id}/chapters/{n}/narrate",
             post(chapter_narrate_fire),
         )
-        .route("/stories/{id}/listen", get(listen_view))
         .route("/stories/{id}/runs", get(runs_view))
         .nest_service("/audio", ServeDir::new(audio_dir))
         .with_state(Arc::new(state))
@@ -201,12 +200,8 @@ async fn new_story_create(
     // parent to compare against. So you get a single first-chapter
     // gen + cleanup pass and status flows to 'complete'.
     if form.fire == "now" {
-        let Ok(database_url) = std::env::var("DATABASE_URL") else {
-            return Err((
-                StatusCode::INTERNAL_SERVER_ERROR,
-                "DATABASE_URL not set — cannot spawn background gen".into(),
-            ));
-        };
+        let database_url = std::env::var("DATABASE_URL")
+            .unwrap_or_else(|_| "postgresql://skald:skald@localhost:5432/skald".into());
         let author_owned = if author_slug.is_empty() {
             None
         } else {
@@ -341,12 +336,7 @@ async fn continue_create(
     // If user clicked "fire now," spawn a background gen task.
     // Otherwise the sequel sits in seed state until CLI fires it.
     if form.fire == "now" {
-        let Ok(database_url) = std::env::var("DATABASE_URL") else {
-            return Err((
-                StatusCode::INTERNAL_SERVER_ERROR,
-                "DATABASE_URL not set — cannot spawn background gen".into(),
-            ));
-        };
+        let database_url = std::env::var("DATABASE_URL").unwrap_or_else(|_| "postgresql://skald:skald@localhost:5432/skald".into());
         let author_owned = if author_slug.is_empty() { None } else { Some(author_slug.to_string()) };
         let direction_owned = direction.clone();
         let chapters = parse_chapters(&form.chapters);
@@ -412,18 +402,7 @@ async fn story_detail(
     .await
     .unwrap_or_default();
 
-    let has_audiobook: bool = sqlx::query_scalar::<_, Option<String>>(
-        "SELECT audiobook_path FROM stories WHERE id = $1",
-    )
-    .bind(id)
-    .fetch_optional(&state.pool)
-    .await
-    .ok()
-    .flatten()
-    .flatten()
-    .is_some();
-
-    let panel = story_panel(&story, &chapters, &characters, &canon_facts, has_audiobook);
+    let panel = story_panel(&story, &chapters, &characters, &canon_facts);
     Ok(Html(render_shell(&stories, Some(id), panel).into_string()))
 }
 
@@ -521,12 +500,8 @@ async fn chapter_narrate_fire(
     let chapter_id =
         chapter_id.ok_or((StatusCode::NOT_FOUND, "chapter not found".into()))?;
 
-    let Ok(database_url) = std::env::var("DATABASE_URL") else {
-        return Err((
-            StatusCode::INTERNAL_SERVER_ERROR,
-            "DATABASE_URL not set — cannot spawn background narrate".into(),
-        ));
-    };
+    let database_url = std::env::var("DATABASE_URL")
+        .unwrap_or_else(|_| "postgresql://skald:skald@localhost:5432/skald".into());
     tokio::spawn(async move {
         if let Err(e) = crate::narrate::run(&database_url, chapter_id, None, 1.0).await {
             tracing::error!(chapter_id = %chapter_id, error = %e, "background narrate failed");
@@ -538,73 +513,6 @@ async fn chapter_narrate_fire(
     Ok(Redirect::to(&format!("/stories/{id}/chapters/{n}")))
 }
 
-/// One chapter as the audiobook player sees it: its start offset
-/// (seconds from the top of the stitched file) and its length.
-#[derive(Debug, Clone)]
-struct AudiobookChapter {
-    n: i32,
-    title: Option<String>,
-    offset_seconds: f64,
-    duration_seconds: f64,
-}
-
-async fn listen_view(
-    State(state): State<Arc<WebState>>,
-    Path(id): Path<Uuid>,
-) -> Result<Html<String>, StatusCode> {
-    let stories = fetch_stories(&state.pool).await;
-    let Some(story) = stories.iter().find(|s| s.id == id).cloned() else {
-        return Err(StatusCode::NOT_FOUND);
-    };
-
-    let audiobook_path: Option<String> = sqlx::query_scalar::<_, Option<String>>(
-        "SELECT audiobook_path FROM stories WHERE id = $1",
-    )
-    .bind(id)
-    .fetch_optional(&state.pool)
-    .await
-    .ok()
-    .flatten()
-    .flatten();
-
-    // Per-chapter durations come from each chapter's most recent
-    // succeeded narration_run. Summed cumulatively in chapter order
-    // they give every chapter's start offset inside the stitched
-    // file — the same order the file was stitched in.
-    let rows: Vec<(i32, Option<String>, Option<f32>)> = sqlx::query_as(
-        r#"
-        SELECT c.n, c.title,
-            (SELECT nr.duration_seconds FROM narration_runs nr
-             WHERE nr.chapter_id = c.id AND nr.status = 'succeeded'
-               AND nr.duration_seconds IS NOT NULL
-             ORDER BY nr.ended_at DESC LIMIT 1)
-        FROM chapters c
-        WHERE c.story_id = $1
-        ORDER BY c.n
-        "#,
-    )
-    .bind(id)
-    .fetch_all(&state.pool)
-    .await
-    .unwrap_or_default();
-
-    let mut chapters = Vec::new();
-    let mut cursor = 0.0_f64;
-    for (n, title, dur) in rows {
-        let d = dur.unwrap_or(0.0) as f64;
-        chapters.push(AudiobookChapter {
-            n,
-            title,
-            offset_seconds: cursor,
-            duration_seconds: d,
-        });
-        cursor += d;
-    }
-
-    let panel = audiobook_panel(&story, audiobook_path.as_deref(), &chapters);
-    Ok(Html(render_shell(&stories, Some(id), panel).into_string()))
-}
-
 async fn runs_view(
     State(state): State<Arc<WebState>>,
     Path(id): Path<Uuid>,
@@ -684,7 +592,7 @@ fn render_shell(stories: &[StoryRow], current: Option<Uuid>, main: Markup) -> Ma
                 }
                 footer.footbar {
                     span { "skald · v0.3 · written down · "
-                        a href="https://git.sulkta.com/cobb/skald" { "cobb/skald" }
+                        a href="http://192.168.0.5:3001/cobb/skald" { "cobb/skald" }
                     }
                 }
             }
@@ -806,7 +714,6 @@ fn story_panel(
     chapters: &[(i32, Option<String>, i32, bool)],
     characters: &[(String, String, String)],
     canon_facts: &[(String, String, String)],
-    has_audiobook: bool,
 ) -> Markup {
     let real_chars: Vec<_> = characters.iter().filter(|c| c.1 == "real").collect();
     let fictional_chars: Vec<_> = characters.iter().filter(|c| c.1 == "fictional").collect();
@@ -824,9 +731,6 @@ fn story_panel(
             }
             nav.story-actions {
                 a.action-primary href=(format!("/stories/{}/continue", s.id)) { "✦ continue this saga" }
-                @if has_audiobook {
-                    a.action-listen href=(format!("/stories/{}/listen", s.id)) { "♪ listen" }
-                }
                 a.action-secondary href=(format!("/stories/{}/runs", s.id)) { "generation log →" }
             }
 
@@ -1073,96 +977,6 @@ fn runs_panel(story_id: Uuid, runs: &[(Uuid, String, String, DateTime<Utc>, Opti
     }
 }
 
-/// Audiobook player — one `<audio>` element over the whole stitched
-/// file plus a clickable chapter list. Clicking a chapter seeks; the
-/// chapter under the playhead highlights as it plays. All client-side
-/// in one small inline script (the file's only JS).
-fn audiobook_panel(
-    story: &StoryRow,
-    audiobook_path: Option<&str>,
-    chapters: &[AudiobookChapter],
-) -> Markup {
-    html! {
-        article.audiobook {
-            a.back href=(format!("/stories/{}", story.id)) { "← back to story" }
-            h1 { (story.title) }
-            (ornament())
-            @match audiobook_path {
-                Some(path) => {
-                    @let basename = path.rsplit('/').next().unwrap_or(path);
-                    @let audio_url = format!("/audio/{}", basename);
-                    @let total: f64 = chapters.iter().map(|c| c.duration_seconds).sum();
-                    p.muted {
-                        (chapters.len()) " chapters · " (fmt_hms(total)) " · "
-                        a href=(audio_url) download=(basename) { "download" }
-                    }
-                    audio #book controls preload="metadata" src=(audio_url) {}
-                    ol.audiobook-chapters {
-                        @for c in chapters {
-                            @let end = c.offset_seconds + c.duration_seconds;
-                            li.ab-chapter
-                                data-seek=(format!("{:.3}", c.offset_seconds))
-                                data-end=(format!("{:.3}", end))
-                            {
-                                span.ab-n { "Chapter " (c.n) }
-                                @if let Some(t) = &c.title {
-                                    span.ab-title { (strip_chapter_prefix(t, c.n)) }
-                                }
-                                span.ab-time { (fmt_hms(c.offset_seconds)) }
-                            }
-                        }
-                    }
-                    script {
-                        (maud::PreEscaped(AUDIOBOOK_JS))
-                    }
-                }
-                None => {
-                    p.muted {
-                        "No audiobook stitched for this saga yet. Render every "
-                        "chapter to audio, then stitch the per-chapter files into one."
-                    }
-                }
-            }
-        }
-    }
-}
-
-const AUDIOBOOK_JS: &str = r#"
-(function () {
-  var audio = document.getElementById('book');
-  if (!audio) return;
-  var rows = Array.prototype.slice.call(document.querySelectorAll('.ab-chapter'));
-  rows.forEach(function (row) {
-    row.addEventListener('click', function () {
-      var t = parseFloat(row.getAttribute('data-seek'));
-      if (!isNaN(t)) { audio.currentTime = t; audio.play(); }
-    });
-  });
-  audio.addEventListener('timeupdate', function () {
-    var now = audio.currentTime;
-    rows.forEach(function (row) {
-      var s = parseFloat(row.getAttribute('data-seek'));
-      var e = parseFloat(row.getAttribute('data-end'));
-      if (now >= s && now < e) { row.classList.add('playing'); }
-      else { row.classList.remove('playing'); }
-    });
-  });
-})();
-"#;
-
-/// Format seconds as H:MM:SS, or M:SS when under an hour.
-fn fmt_hms(s: f64) -> String {
-    let total = s.max(0.0) as i64;
-    let h = total / 3600;
-    let m = (total % 3600) / 60;
-    let sec = total % 60;
-    if h > 0 {
-        format!("{h}:{m:02}:{sec:02}")
-    } else {
-        format!("{m}:{sec:02}")
-    }
-}
-
 // ─── helpers ─────────────────────────────────────────────────────
 
 /// Knotwork divider — a small SVG ornament used as section break.
@@ -1504,45 +1318,6 @@ code { font-family: var(--mono); font-size: 0.9em; background: var(--surface-2);
 
 .empty { color: var(--ink-faint); font-style: italic; }
 
-/* ─── audiobook player ─────────────────────────────────────── */
-.action-listen {
-    font-family: var(--display); letter-spacing: 2px; font-size: 12px;
-    text-transform: uppercase; color: var(--bronze);
-    border: 1px solid var(--bronze-dim); padding: 9px 18px;
-    background: var(--surface);
-}
-.action-listen:hover { color: var(--bg); background: var(--bronze); border-color: var(--bronze); }
-.audiobook .back {
-    display: inline-block; color: var(--ink-faint); font-size: 12px;
-    margin-bottom: 18px; letter-spacing: 0.5px;
-}
-.audiobook h1 {
-    font-family: var(--display); font-size: 32px; color: var(--ink);
-    margin: 0; font-weight: 700; letter-spacing: 1px;
-}
-.audiobook > .muted { font-size: 13px; margin: 0 0 14px 0; }
-.audiobook > .muted a { color: var(--bronze); }
-.audiobook audio { width: 100%; margin: 4px 0 26px 0; }
-.audiobook-chapters { list-style: none; margin: 0; padding: 0; }
-.ab-chapter {
-    display: grid; grid-template-columns: 120px 1fr auto; gap: 20px;
-    align-items: baseline; padding: 13px 16px; cursor: pointer;
-    border-left: 2px solid transparent; border-bottom: 1px solid var(--surface-2);
-    transition: background 80ms ease;
-}
-.ab-chapter:hover { background: var(--surface-2); border-left-color: var(--bronze-dim); }
-.ab-chapter.playing { background: var(--surface-2); border-left-color: var(--accent); }
-.ab-n {
-    font-family: var(--mono); font-size: 12px; color: var(--bronze-dim);
-    letter-spacing: 1px;
-}
-.ab-chapter.playing .ab-n { color: var(--accent); }
-.ab-title { font-family: var(--serif); font-size: 16px; color: var(--ink); }
-.ab-time {
-    font-family: var(--mono); font-size: 12px; color: var(--ink-faint);
-    text-align: right;
-}
-
 /* ─── forms (new-saga + continue) ──────────────────────────── */
 .form-panel h1 {
     font-family: var(--display); font-size: 30px; color: var(--ink);
@@ -1604,7 +1379,6 @@ code { font-family: var(--mono); font-size: 0.9em; background: var(--surface-2);
     .topnav { margin-left: 0; width: 100%; padding-top: 6px; }
     .chapter-list a { grid-template-columns: 80px 1fr auto; }
     .chapter-list .wc { display: none; }
-    .ab-chapter { grid-template-columns: 80px 1fr auto; gap: 12px; }
     .char-list li { grid-template-columns: 1fr; gap: 4px; }
     .cname { font-size: 14px; }
     .brand { font-size: 22px; letter-spacing: 3px; }