@event4u/agent-config 5.4.1 → 5.6.0

package/.agent-src/skills/image-analyser/canon-spec.md

@@ -0,0 +1,109 @@
+# Character Canon Spec — schema, fidelity rubric, fidelity loop
+
+Shared contract consumed by [`image-analyser`](SKILL.md) and
+[`image-creator`](../image-creator/SKILL.md). Defines the structured truth a
+character image is reconciled against, how a candidate is scored, and how the
+create→analyse→regenerate loop converges.
+
+> **Design lock (AI-council, anthropic/claude-sonnet-4-5 + openai/gpt-4o, 2-round
+> debate, user-invoked):** keep **ontology and epistemology separate**.
+> `confidence` is a property of a *verification attempt*, **not** of the
+> character — so it never lives on a canon leaf. Three layers below. The rubric
+> is a **vector + hard gate**, never one scalar. The loop uses **plateau +
+> oscillation detection**, not a bare iteration count.
+
+## The three layers
+
+A character record is split so each layer changes for a different reason:
+
+### Layer 1 — `identity` (immutable canon · the character truth)
+
+What the character *is*. Per-leaf `severity` only — **no confidence**, no
+verification state. Source of truth = the canon (the book + its authoritative
+portraits; *the image wins over the text*).
+
+```jsonc
+{
+  "id": "veikko",
+  "identity": {
+    "physique": { "value": "lean wiry athletic, 1.82m, broad shoulders narrow hips", "severity": "major" },
+    "face":     { "value": "...", "marks": [ { "value": "tiny scar above right mouth corner", "severity": "minor" } ] },
+    "hair":     { "value": "vertical split: LEFT pitch-black / RIGHT platinum-blond, long open to chest, no shaved sides", "severity": "canon-breaking" },
+    "eyes":     { "left": "ice-blue", "right": "forest-green", "heterochromia": true, "severity": "canon-breaking" },
+    "tattoos":  [ { "location": "central chest", "motif": "Vegvisir compass", "style": "blackwork", "severity": "canon-breaking" },
+                  { "location": "left chest", "motif": "Loki serpent biting tail", "severity": "canon-breaking" },
+                  { "location": "knuckles", "motif": "block letters", "text": "S-U-S-I", "severity": "major" } ],
+    "jewelry":  [ { "value": "massive round silver watch, RIGHT wrist", "severity": "major" } ],
+    "outfit_variants": [ { "name": "studio-casual", "value": "black sleeveless tank under open leather vest" } ]
+  },
+  "identity_anchors": ["hair", "eyes", "tattoos[central chest]", "tattoos[left chest]", "jewelry[watch]"],
+  "notes": "Loki = asymmetry everywhere (hair, eyes, tattoos)."
+}
+```
+
+- **`severity`** ∈ `canon-breaking | major | minor` — how much a miss matters.
+- **`identity_anchors`** — the must-never-drift list. **Derived rule:** every
+  `severity: canon-breaking` leaf MUST be an anchor; anchors MAY also name
+  cross-feature invariants (e.g. "asymmetry") that no single leaf captures.
+- Relationship to [`character-consistency`](../character-consistency/SKILL.md):
+  its existing token JSON (`agents/reference/ai-video/<project>/characters/<id>.json`)
+  is the **load-bearing subset** of this `identity` layer. The Canon Spec is the
+  richer superset; `image-analyser` emits the token subset into that exact file
+  so there is **one** character record, not two.
+
+### Layer 2 — `observation` (verification state · the analyser's output)
+
+What a *specific image* shows, per attempt. **This** is where `confidence`
+lives (`high | medium | low`, the `image-ocr` pattern) plus `unverifiable[]`
+for features the image cannot resolve (occluded / low-res). Never written back
+onto Layer 1.
+
+```jsonc
+{
+  "source": "agents/tmp/odins-beard/img_2.png", "character": "veikko",
+  "observed": { "hair": { "value": "near-uniform light/blond, split not distinct", "confidence": "high" },
+                "eyes": { "value": "both read blue; green not visible", "confidence": "medium" } },
+  "unverifiable": ["tattoos[knuckles].text (hands out of frame)"]
+}
+```
+
+### Layer 3 — `generative_hints` (prompt-assembly guidance)
+
+How to render the identity well: anchor ordering (hard-to-render anchors first
+— heterochromia, hair-split), per-engine caveats, negative-prompt seeds. Read
+by `image-creator`; never confused with the canon itself.
+
+## Fidelity rubric — vector + hard gate (not one scalar)
+
+A diff scores each observed feature `match | partial | miss`, then reports a
+**vector**, not a single number:
+
+1. **Canon-breaking gate (hard).** ANY `canon-breaking` leaf at `miss` → overall
+   **FAIL**, regardless of everything else. Non-negotiable.
+2. **Per-section scores.** `face`, `hair`, `eyes`, `tattoos`, `outfit`, `jewelry`
+   each get their own 0–100 (severity-weighted within the section). Surfaced
+   individually so a strong face can't mask a broken hair split.
+3. **Headline.** A weighted roll-up is shown for convenience but is **advisory** —
+   the gate + per-section vector decide pass/fail, not the roll-up.
+4. **Low-confidence discipline.** A `miss` on a `low`-confidence observation is
+   reported as `needs-better-image`, **not** counted as a hard miss — avoids
+   false-fail on an un-resolvable feature (re-pass per SKILL § enhancement first).
+
+## Fidelity loop — plateau + oscillation detection
+
+`image-creator` generates → `image-analyser` re-reads the output against the
+character's Layer-1 identity → diff → feed `canon-breaking` + `major` misses back
+as refined prompt directives → regenerate.
+
+Stop conditions (first to fire):
+
+- **PASS** — canon-breaking gate clear AND every per-section score ≥ its
+  threshold.
+- **Plateau** — N consecutive rounds with no per-section score improvement →
+  stop; the prompt is not the bottleneck (provider/seed is).
+- **Oscillation** — fixing feature X regresses a previously-passing feature Y
+  (tracked across rounds) → stop; surface the trade-off, do not thrash.
+- **Budget** — hard ceiling on rounds as a backstop.
+
+On any non-PASS stop → **halt and surface** the best candidate + its remaining
+diff for human review. **Never silently accept drift** (per `verify-before-complete`).

package/.agent-src/skills/image-analyser/evals/triggers.json

@@ -0,0 +1,16 @@
+{
+  "skill": "image-analyser",
+  "description": "5 should-trigger + 5 should-not-trigger. Should-trigger covers the analyse / canon-diff / accuracy-check / drift-find / bootstrap-from-portrait paths (DE + EN). Should-not covers the near-miss neighbours whose vocabulary overlaps: scene/motion review (video-director), non-character art (canvas-design), cross-scene token lock (character-consistency), generation (image-creator), and document OCR (generic).",
+  "queries": [
+    {"q": "analyse this character image down to the last detail", "trigger": true},
+    {"q": "does img_2.png match the Charakterbuch canon?", "trigger": true},
+    {"q": "check this render for character accuracy — what drifted?", "trigger": true},
+    {"q": "vergleiche das Bild mit der Charakterbeschreibung, jedes Merkmal", "trigger": true},
+    {"q": "build a canon spec from this authoritative portrait", "trigger": true},
+    {"q": "turn this scene idea into a cinematic video prompt", "trigger": false, "note": "scene/motion → video-director"},
+    {"q": "design a poster for the band", "trigger": false, "note": "static non-character art → canvas-design"},
+    {"q": "lock this character's tokens so every scene reuses them", "trigger": false, "note": "cross-scene token lock → character-consistency"},
+    {"q": "generate Veikko in the forge scene to spec", "trigger": false, "note": "generation → image-creator"},
+    {"q": "extract the text from this scanned invoice", "trigger": false, "note": "plain document OCR, not character analysis"}
+  ]
+}

package/.agent-src/skills/image-creator/SKILL.md

@@ -0,0 +1,117 @@
+---
+model_tier: high
+name: image-creator
+description: "Use to generate a character image to spec — max-fidelity reproducible prompt from a Canon Spec, anchors-first, provider/governance-gated. Triggers 'generate this character', 'render to spec'."
+personas:
+  - hollywood-director
+domain: product
+workspaces:
+  - small-business
+packs:
+  - ai-video
+lifecycle: experimental
+trust:
+  level: experimental
+install:
+  default: false
+  removable: true
+---
+
+# image-creator
+
+> Turn a **Canon Spec** + a scene into a maximally-detailed, reproducible
+> generation prompt that renders a character to spec — then hand the result to
+> [`image-analyser`](../image-analyser/SKILL.md) to verify. The loop partner.
+> Schema + rubric + loop: [`canon-spec.md`](../image-analyser/canon-spec.md).
+
+## When to use
+
+- "Generate / render / create this character", "render Veikko in scene X to
+  spec", "make the image match the canon".
+- Inside the fidelity loop, fed by `image-analyser`'s correction directives.
+
+NOT for: scene blocking / motion (→ `video-director`, `motion-choreographer`,
+which take a verified still from here), non-character art (→ `canvas-design`).
+
+## Input
+
+- Character id / Canon Spec (`agents/reference/ai-video/<project>/characters/<id>.json`).
+- Scene brief (setting + pose). Optional: prior `image-analyser` diff (loop mode).
+
+## Procedure
+
+1. **Governance gate FIRST** (per `media-governance-routing`): if the character
+   is a real-person likeness, consult `agents/settings/policies/media/likeness.md`
+   + `public-figures.md` + `disclosure.md` before emitting anything. Fictional
+   characters (odins-beard trio) are exempt; the routing decision is in-session.
+2. **Provider gate** (per `provider-lifecycle-discipline`): read the resolved
+   provider's tier; if non-stable (experimental/deprecated/community), surface
+   the tier and ask before running. Never default to a non-stable provider
+   silently. Name the provider + tier in the run summary.
+3. **Assemble the prompt from the spec — anchors first.** Order matters: the
+   hard-to-render `identity_anchors` go at the TOP (the canon's lesson —
+   heterochromia + hair-split get forgotten if buried). Then physique, face
+   (+ marks), per-location tattoos (incl. exact `text`), outfit, jewelry.
+4. **Asymmetry block** — for split / heterochromatic characters, an explicit
+   left/right section ("LEFT half black / RIGHT half blond", "blue LEFT eye /
+   green RIGHT eye") with concrete comparison refs for engines that drop it.
+5. **Negative block + engine settings** — reuse the canon's proven structure
+   (negatives that kill "single hair colour", "both eyes same colour", etc.;
+   per-engine `--ar`/`--style`/CFG/steps). Do not reinvent; the character book's
+   prompt format is the template.
+6. **Generate** through the existing provider/adapter layer
+   (`scripts/ai-video/adapters/`, the `/video|image` surface). Do **not** add a
+   new provider path where one exists.
+7. **Verify** — hand the output to `image-analyser`; in loop mode, fold its
+   correction directives into the next prompt (see the loop in `canon-spec.md`).
+
+## Anchors-first — why
+
+The fidelity-loss evidence (`img_2.png`): Veikko's split hair + heterochromia
+were missed because they are hard for the engine and were not front-loaded. The
+analyser-derived fixes go to the TOP of the next prompt, with negatives, and the
+loop re-verifies — that is how the smallest mole comes back.
+
+## Output format
+
+1. **Generation prompt** — anchors block · positive · asymmetry (if any) · negative · engine settings.
+2. **Provider + tier line** (the audit entry).
+3. **Verify call** — the `image-analyser` invocation on the result + the loop stop-state.
+
+## Example (anchors-first vs buried)
+
+- Safe (Veikko): prompt opens with `HAIR: exact vertical centre split, LEFT pitch-black, RIGHT platinum-blond` + `EYES: heterochromia — LEFT ice-blue, RIGHT forest-green`, then the rest.
+- Unsafe: physique/outfit first, the split + heterochromia in a trailing sentence → the engine drops them (the observed `img_2.png` failure).
+
+## Gotchas
+
+- Never claim "canon-perfect" without an `image-analyser` pass (per `verify-before-complete`).
+- DALL-E drops heterochromia + split hair — use concrete comparisons; expect multiple generations.
+- Keep the spec the single source: regenerate from the Canon Spec, never paraphrase it into a fresh prompt.
+
+## Do NOT
+
+- Do NOT claim "canon-perfect" / "matches the canon" without an `image-analyser`
+  pass on the output (per `verify-before-complete` — no completion without evidence).
+- Do NOT bury the `identity_anchors` — the hard-to-render features (heterochromia,
+  hair-split) go at the TOP of the prompt, never in a trailing sentence.
+- Do NOT default to a non-stable provider silently — surface the tier and ask
+  first (per `provider-lifecycle-discipline`).
+- Do NOT emit a prompt for a real-person likeness before the governance gate
+  (step 1) clears it.
+- Do NOT add a new provider/generation path where the existing adapter layer
+  (`scripts/ai-video/adapters/`) already covers it.
+- Do NOT paraphrase the Canon Spec into a fresh prompt — regenerate from the spec
+  so it stays the single source of truth.
+
+## Policies
+
+`media-governance-routing` + `agents/settings/policies/media/` (likeness, style,
+public-figures, disclosure) — consulted in step 1 before any prompt is emitted
+for a real-person likeness.
+
+## Related skills
+
+- [`image-analyser`](../image-analyser/SKILL.md) — the verify/loop partner.
+- [`character-consistency`](../character-consistency/SKILL.md) — supplies the locked identity tokens.
+- [`video-director`](../video-director/SKILL.md) / [`motion-choreographer`](../motion-choreographer/SKILL.md) — take the verified still into scene + motion.

package/.agent-src/skills/image-creator/evals/triggers.json

@@ -0,0 +1,16 @@
+{
+  "skill": "image-creator",
+  "description": "5 should-trigger + 5 should-not-trigger. Should-trigger covers generate/render/to-spec/loop-fix paths (DE + EN). Should-not covers near-miss neighbours: analysis (image-analyser), scene/motion (video-director, motion-choreographer), non-character art (canvas-design), and token-lock (character-consistency).",
+  "queries": [
+    {"q": "generate Veikko in the forge scene, matched to the canon", "trigger": true},
+    {"q": "render this character to spec, max detail", "trigger": true},
+    {"q": "create the image and make every tattoo placement exact", "trigger": true},
+    {"q": "erzeuge das Bild so detailgenau wie möglich nach der Vorlage", "trigger": true},
+    {"q": "regenerate with the heterochromia and hair-split fixed", "trigger": true},
+    {"q": "analyse this image and tell me what drifted", "trigger": false, "note": "analysis → image-analyser"},
+    {"q": "turn this still into a cinematic motion prompt for Veo", "trigger": false, "note": "motion → motion-choreographer"},
+    {"q": "expand this one-line idea into a scene blueprint", "trigger": false, "note": "scene → scene-expander/video-director"},
+    {"q": "design a tour poster", "trigger": false, "note": "static non-character art → canvas-design"},
+    {"q": "lock the character tokens so all scenes reuse them", "trigger": false, "note": "token lock → character-consistency"}
+  ]
+}

package/.agent-src/skills/song-to-script/SKILL.md

@@ -0,0 +1,216 @@
+---
+model_tier: inherit
+name: song-to-script
+description: "Turn an audio track into a timed `## Scene N` script: song sections → per-scene durations, auto mode adds mood + lip-sync lines. Triggers 'music video', 'from the song', 'cut to the beat'."
+personas:
+  - hollywood-director
+domain: product
+workspaces:
+  - small-business
+packs:
+  - ai-video
+lifecycle: experimental
+trust:
+  level: experimental
+install:
+  default: false
+  removable: true
+---
+
+# song-to-script
+
+> Turn a song into `<project>/script.md` — a sequence of `## Scene N`
+> blocks whose `duration:` values sum to the track length and whose cut
+> points land on real section boundaries. Consumed by
+> [`/video:from-song`](../../commands/video/from-song.md), then handed
+> to [`scene-expander`](../scene-expander/SKILL.md) and
+> [`video-director`](../video-director/SKILL.md). Never invents timing —
+> every boundary comes from the audio probe, and the probe's `method`
+> tells this skill how musical (or not) those boundaries actually are.
+
+## When to use
+
+- A music-video run needs scenes cut to the song (`/video:from-song`).
+- An existing script must be **re-timed** to a track after the edit
+  drifted from the beat (the named second consumer — re-time without a
+  full re-author).
+
+Do NOT use when:
+
+- The operator already supplies a `## Scene N` script with `duration:`
+  values — feed it straight to `scene-expander`.
+- There is no audio — use the operator brief with `scene-expander`
+  directly.
+
+## Inputs
+
+- **Audio probe** — JSON from
+  [`scripts/ai-video/lib/probe-audio.sh`](../../../../scripts/ai-video/lib/probe-audio.sh):
+  `{duration, method, warning?, sections:[{start,end,energy,label}]}`.
+  - `method: silence` — boundaries are real quiet gaps; trust them as cuts.
+  - `method: rms` — boundaries are energy-delta inflections; usable but
+    coarse.
+  - `method: interval` — **the track is structurally flat** (brick-walled
+    / sustained); sections are fixed-interval, NOT musical. When `method`
+    is `interval` (or `warning` is set), the emitted script header states
+    that timing is interval-based and the operator should pass
+    `--scene-durations` for musical sync. Never present interval cuts as
+    beat-synced.
+- **Mode** — `brief` (operator text is the creative source) or `auto`
+  (infer mood + action from energy).
+- **Brief** (brief mode only) — free text: story, settings, look.
+- **Character lock** (optional) — `<project>/character.json` if a human
+  subject was locked. **Absent is normal** — abstract / landscape /
+  visualiser videos have no locked subject; see Step 2.
+
+## Procedure
+
+### Step 1: Map sections → scenes
+
+One `## Scene N` per probe section. `duration:` = `end - start`
+(rounded to 0.5 s). Merge any section shorter than the provider's
+`min-duration` into its neighbour; split any section longer than the
+provider's `max-duration` into equal sub-scenes so no single clip
+exceeds the backend limit (read both from the resolved provider tuning).
+
+### Step 2: Assign mood + action
+
+First decide the **subject mode**:
+
+- **Character mode** — `character.json` exists: every scene's `action:`
+  names the locked subject, never a fresh description.
+- **Style mode** — no `character.json`: scenes describe setting, palette,
+  and motion continuity (the recurring *look*), not a person. This is the
+  valid abstract / landscape / visualiser path — do not invent a human
+  subject to fill the slot.
+
+Then assign per scene:
+
+- **Brief mode** — distribute the brief's beats across scenes in order;
+  energy only modulates pacing (low energy → slow push-in, high energy →
+  fast cuts / motion). Do not add story the brief did not state.
+- **Auto mode** — derive mood per section from `energy` and `label`:
+
+  | label / energy | default scene intent |
+  |---|---|
+  | intro / low | establishing wide, slow camera, calm subject/scene |
+  | build / rising | approach, tightening framing |
+  | drop / peak | dynamic motion, weather/FX, fast push |
+  | breakdown / dip | close-up / detail, quiet, single light source |
+  | outro / fade | pull-back, resolve, hold |
+
+### Step 3: Vocal map — transcribe, never guess (vocal tracks)
+
+```
+LYRIC TIMING AND SINGER COME FROM THE TRANSCRIBED AUDIO, NEVER FROM A
+BRIEF / STORY SKELETON OR A GUESSED STRETCH. NEVER PUT ONE SINGER'S
+LINE ON ANOTHER SINGER'S SCENE.
+```
+
+Track has vocals and the run intends lip-sync → build a **vocal map**
+from the real audio before assigning any `dialogue:`:
+
+1. **Transcribe** audio to timestamped lines — OpenAI
+   `/v1/audio/transcriptions` (`response_format=verbose_json` →
+   `segments[].{start,end,text}`) or local whisper. Only source of lyric
+   timing.
+2. **Label the singer** per line. Operator who-sings reference (roster, a
+   brief naming who sings which line, or a character cast) → match each
+   line to its singer. Genuinely ambiguous → mark `singer: "?"` and
+   surface; never guess a singer to fill the slot.
+3. **Emit** `<project>/vocal-map.json`: `[{start, end, text, singer}]`,
+   timing verbatim from the transcript.
+4. **Place lines into the matching scene's** `dialogue:` block using
+   transcript timing, tagged with the singer (`singer: "<line>"`). A
+   scene's lip-sync subject MUST be the line's labelled singer.
+
+No vocals / no transcript / no lip-sync intent → leave `dialogue:` empty;
+scene is performance / B-roll. **Never fabricate lyrics**, **never
+re-time a line off the brief**, and in style mode `dialogue:` stays empty
+(lip-sync needs a character subject). The `/video:from-song` sign-off
+gate (its Step 6) shows this map for approval before any render.
+
+### Step 4: Emit + reconcile
+
+Write `<project>/script.md` (and `<project>/vocal-map.json` when the
+track has vocals). Report the delta, the section→scene map, **the probe
+`method`** (silence-derived, energy-derived, or interval-fallback), **and
+whether lyric timing is transcript-derived** (it must be — never
+brief-derived). If the sum cannot be reconciled (e.g. provider
+max-duration forces more time than the song has), **halt and surface the
+conflict** — do not pad silently.
+
+### Step 5: Validate before handoff
+
+Concrete checks (all must pass before the script is handed to
+`scene-expander`):
+
+- **Assert** `Σ(duration) == probe.duration` within ±1.0 s; report the
+  exact delta. A larger delta → halt, do not pad.
+- **Verify** every scene boundary equals a probe section boundary (or a
+  `--scene-durations` value) — no invented cut points.
+- **Confirm** no scene `duration:` exceeds the provider `max-duration`
+  or falls below `min-duration`.
+- **Ensure** every `## Scene N` carries all five keys (`duration` ·
+  `mood` · `action` · `camera` · `dialogue`), and that `dialogue:` is
+  empty in style mode.
+
+## Output format
+
+```markdown
+# <project> — derived from <song-file> (<mode> mode · cuts: <method>)
+
+## Scene 1
+duration: 6.0
+mood: establishing, cold, pre-storm
+action: <subject from character.json, OR style description in style mode>
+camera: slow push-in
+dialogue:
+
+## Scene 2
+duration: 4.5
+mood: build, rising tension
+action: close on <subject / detail>, wind picking up
+camera: handheld tighten
+dialogue:
+  - "<subject>: \"<lyric line for this section, if any>\""
+```
+
+`scene-expander` consumes this verbatim — keep the keys
+(`duration` · `mood` · `action` · `camera` · `dialogue`) exact.
+
+## Gotcha
+
+- **`method: interval` is the brick-walled-master signal, not a bug.**
+  A compressed modern master has near-constant RMS and no silence, so the
+  probe degrades to fixed intervals. That is the honest floor — surface
+  it and point the operator at `--scene-durations`; never dress interval
+  cuts up as beat-synced.
+- **A vocal section without supplied lyrics is B-roll, not lip-sync.**
+  Detected vocal energy alone does not authorise `dialogue:` — only
+  operator-supplied lyrics do.
+- **Style mode is the default for a no-character run**, not an error
+  path. Landscape / abstract / visualiser videos never get a fabricated
+  human subject.
+
+## Do NOT
+
+- **Do NOT invent timing.** Every cut maps to a probe boundary or a
+  `--scene-durations` value — never to taste.
+- **Do NOT present `interval`-fallback cuts as beat-synced.** Always
+  surface the probe `method`.
+- **Do NOT emit a clip outside the provider's min/max duration** —
+  split/merge in Step 1 instead.
+- **Do NOT fabricate lyrics or story** beyond the brief / detected vocals.
+- **Do NOT invent a human subject** in style mode; defer identity to
+  `character.json` only when a lock exists.
+- **Do NOT pad a unreconcilable timing sum** — halt and surface it.
+
+## See also
+
+- [`/video:from-song`](../../commands/video/from-song.md) — the command
+  that drives this skill
+- [`scene-expander`](../scene-expander/SKILL.md) — consumes the emitted
+  script
+- [`character-consistency`](../character-consistency/SKILL.md) — supplies
+  the locked subject referenced in `action:` (character mode only)

package/.claude-plugin/marketplace.json

@@ -6,7 +6,7 @@
   },
   "metadata": {
     "description": "Shared agent configuration \u2014 skills for AI coding tools (Claude Code, Augment, Cursor, Cline, Windsurf, Gemini CLI).",
-    "version": "5.4.1",
+    "version": "5.6.0",
     "keywords": [
       "agent-config",
       "skills",
@@ -187,6 +187,12 @@
         "./.claude/skills/gtm-launch",
         "./.claude/skills/guideline-writing",
         "./.claude/skills/hiring-loop-design",
+        "./.claude/skills/image",
+        "./.claude/skills/image-analyse",
+        "./.claude/skills/image-analyser",
+        "./.claude/skills/image-create",
+        "./.claude/skills/image-creator",
+        "./.claude/skills/image-verify",
         "./.claude/skills/implement-ticket",
         "./.claude/skills/incident-commander",
         "./.claude/skills/jira-integration",
@@ -349,6 +355,7 @@
         "./.claude/skills/skill-management",
         "./.claude/skills/skill-reviewer",
         "./.claude/skills/skill-writing",
+        "./.claude/skills/song-to-script",
         "./.claude/skills/sql-writing",
         "./.claude/skills/stakeholder-tradeoff",
         "./.claude/skills/subagent-orchestration",
@@ -384,13 +391,19 @@
         "./.claude/skills/video",
         "./.claude/skills/video-director",
         "./.claude/skills/video-from-script",
+        "./.claude/skills/video-from-song",
         "./.claude/skills/video-scene",
         "./.claude/skills/video-stitch",
         "./.claude/skills/video-storyboard",
         "./.claude/skills/vision-articulation",
         "./.claude/skills/voc-extract",
         "./.claude/skills/voice-and-tone-design",
-        "./.claude/skills/work"
+        "./.claude/skills/work",
+        "./.claude/skills/knowledge-cross-repo",
+        "./.claude/skills/skill",
+        "./.claude/skills/skill-preview",
+        "./.claude/skills/skills",
+        "./.claude/skills/skills-discover"
       ]
     }
   ]

package/CHANGELOG.md

@@ -811,6 +811,90 @@ our recommendation order, not its support status.
 > that forces a new era split (`# Era: 5.5.x`, etc.) — see
 > [`docs/contracts/CHANGELOG-conventions.md § Era splits`](docs/contracts/CHANGELOG-conventions.md).
 
+## [5.6.0](https://github.com/event4u-app/agent-config/compare/5.5.0...5.6.0) (2026-05-31)
+
+### Features
+
+* **ai-video:** add /image command cluster (analyse/create/verify) ([9d53c81](https://github.com/event4u-app/agent-config/commit/9d53c81e3aed3549c4135a641e306a32fa224a68))
+* **ai-video:** add image-analyser + image-creator character-fidelity skills ([b76a4db](https://github.com/event4u-app/agent-config/commit/b76a4db19d537ad5443f2c920c4701a49d08e584))
+* **ai-video:** extend from-song + song-to-script + adapters; add media-sync-ground-truth rule ([dab9f0f](https://github.com/event4u-app/agent-config/commit/dab9f0f41e0d4b93eef423541269d54408bf36f4))
+* **kernel:** kernel-budget soak — trim commit-policy + scope-control, fences byte-identical (Phase 1) ([b0547f0](https://github.com/event4u-app/agent-config/commit/b0547f0207df6e057e2ff97283391ddad88fd2e3))
+* **schema:** tighten description caps + warning window (Phase 2.1); record 3.2 as obviated ([c65d3ae](https://github.com/event4u-app/agent-config/commit/c65d3ae8b569eb3507b83c31967231e64a31a742))
+* **projection:** thin rule-layer projection behind a flag — measured -35,845 GPT tok (Phase 3.1) ([d786363](https://github.com/event4u-app/agent-config/commit/d7863635ee342ab61210ed14b95eeb1fface9582))
+* **value:** drop the € comparison from the value dashboard — tokens only ([33f6fc8](https://github.com/event4u-app/agent-config/commit/33f6fc81e934ca426e5a4c82c37f41624228a636))
+* **bench:** drop the monetary cost comparison — report tokens only ([1a2d3ad](https://github.com/event4u-app/agent-config/commit/1a2d3ad6a177122f0249bb903b3d7d74d62676e1))
+* **audit:** unified audit:tokens analyzer + budget CI gate (Phase 0B.2/0B.4/1.3/1.4) ([c27061f](https://github.com/event4u-app/agent-config/commit/c27061fd47be600d79a5b0a8484a4061f9cdadfd))
+* **safety:** trigger-coverage MUST-LOAD floor + thin-projection kill-switch (Phase 2) ([b210e0b](https://github.com/event4u-app/agent-config/commit/b210e0b9f402e26c813d6fd45e85fdad42876c1d))
+* **budget:** add real-tokenizer measurement alongside chars (Phase 0B.1) ([1602d7f](https://github.com/event4u-app/agent-config/commit/1602d7f2f41a3d6b1069ad675583262542c970cf))
+
+### Bug Fixes
+
+* **ai-video:** add workspaces + packs to media-sync-ground-truth rule ([cebe846](https://github.com/event4u-app/agent-config/commit/cebe8465d5cc1b07c8426b91ab52c0fb07e79f5b))
+* sync command count to 145 in README badge + browse lines ([7a881e2](https://github.com/event4u-app/agent-config/commit/7a881e243fe73faad725050d684af682445c4eda))
+* **kernel:** restore 2nd 'Iron Law' mention in commit-policy (obligation-baseline regression from #310) ([693c287](https://github.com/event4u-app/agent-config/commit/693c28791cbe124942835c3b309ca72921ee13f5))
+* **tools:** repoint audit_auto_rules + audit_command_surface to packages/core (Phase 2.3) ([fae234c](https://github.com/event4u-app/agent-config/commit/fae234cfda21372adb493b259a47e8a9a00a0036))
+* **schema:** rule description cap 160 -> 190 (smoke gate treats over-cap as FAIL) ([72f05e0](https://github.com/event4u-app/agent-config/commit/72f05e02cca4e196e514d8f4f573ac9890f52583))
+* **kernel:** repoint iron_law_sha to packages/core layout (Iron-Law SHA gate was broken) ([aab5755](https://github.com/event4u-app/agent-config/commit/aab57558f7047fc21d0107dbab5aa3b49ebe4e2f))
+* **schema:** add lean_projection.mode to the settings schema (template-parity gate) ([51ed03e](https://github.com/event4u-app/agent-config/commit/51ed03ec9e74d4573b7da43a6b45804535986690))
+* **bench:** repoint bench_runner SKILLS_DIR to .agent-src for packages/core layout ([e183527](https://github.com/event4u-app/agent-config/commit/e183527c0efa4e3fe7dfb6a9df2c1e21534925a8))
+
+### Performance
+
+* **projection:** minimal thin entries — measured saving 35,845 -> 45,182 GPT tok (77%) ([b1aabe0](https://github.com/event4u-app/agent-config/commit/b1aabe05a9e68aa87a8396f3276530d480582dbe))
+
+### Documentation
+
+* **roadmap:** honest dispositions for remaining build-out items ([68e8bb5](https://github.com/event4u-app/agent-config/commit/68e8bb5ab3ec95c516707d9b8e0f2bfd6f7c5ec9))
+* **roadmap:** land Phase 0+2 of road-to-lean-initial-context; spawn build-out follow-up ([8b0bec2](https://github.com/event4u-app/agent-config/commit/8b0bec2efab7128fa4c45d186e3feea4ebd632f8))
+
+### Chores
+
+* update reports ([eb52977](https://github.com/event4u-app/agent-config/commit/eb52977cf398f6691e55fc063ba9f1b4f63f03b9))
+* regenerate index + catalog for image skills + /image cluster ([f71f024](https://github.com/event4u-app/agent-config/commit/f71f024dba491ff36c071d9fd25d372e39b0c9c2))
+* regenerate derived (router, marketplace, manifests, counts, command-surface) ([2fcc321](https://github.com/event4u-app/agent-config/commit/2fcc3214451bec3114dee0d55d7f03c0f3392ca0))
+* archive road-to-character-image-fidelity (complete) ([ef16067](https://github.com/event4u-app/agent-config/commit/ef16067c0636a0c8279922bbb160531efb477ddb))
+* track linter-debt roadmap + command-surface report (unrelated to lean-context) ([a52a604](https://github.com/event4u-app/agent-config/commit/a52a6048a55805e5a25b9dc68fd1e98091ba0fec))
+* **router:** regenerate stale dist/router.json from source ([af782fe](https://github.com/event4u-app/agent-config/commit/af782fe179d7ba573ec916e48ac0b9e3752cf6f5))
+
+Tests: 5375 (+39 since 5.5.0)
+
+## [5.5.0](https://github.com/event4u-app/agent-config/compare/5.4.1...5.5.0) (2026-05-31)
+
+### Features
+
+* /skill:preview — non-destructive skill dry-run ([03ce5fc](https://github.com/event4u-app/agent-config/commit/03ce5fc757f9b42226083627ad7380dd0d895ef7))
+* cross-repo retrieval + linked-projects:list (ADR-032 Option A) ([894e2e8](https://github.com/event4u-app/agent-config/commit/894e2e88c2a7e6744eb45a90907c634953aec0ec))
+* /skills:discover — local, explained skill recommender ([d887c56](https://github.com/event4u-app/agent-config/commit/d887c561514b019f2ae1a3eca068087f4ffee351))
+* meta-layer concept-surface audit tool + zero-cut evidence ([b4a7f3e](https://github.com/event4u-app/agent-config/commit/b4a7f3e5e87ab0c5aa4a820784fba09d8306284a))
+* **video:** add /video:from-song music-video command + register in cluster ([05752df](https://github.com/event4u-app/agent-config/commit/05752dfd0837759923bf02f3731fecfb81dec9e7))
+* **ai-video:** add probe-audio.sh hybrid audio segmentation ([41b4f8d](https://github.com/event4u-app/agent-config/commit/41b4f8dd57d1352df49bdc31bc81ea8e7e04f550))
+
+### Bug Fixes
+
+* classify command-cluster files as commands, not skills ([0fd25da](https://github.com/event4u-app/agent-config/commit/0fd25dad8f75cfb7021512e2905b23b24fea88c1))
+* **ai-video:** portability + regenerate stale derived for from-song ([9d728b8](https://github.com/event4u-app/agent-config/commit/9d728b88687a6c27dfa647dd233fb4412f7e8d5a))
+
+### Documentation
+
+* reword execution-type mentions to dodge check-refs false positive ([dc84ed0](https://github.com/event4u-app/agent-config/commit/dc84ed01d3aa6845a1e15674ed1ec3ff3bbfb9ce))
+* add discoverable BREAKING_CHANGES.md + major-bump rationale ([385b8a2](https://github.com/event4u-app/agent-config/commit/385b8a225b764cfe09a7e31640184140de4e07d9))
+* **roadmap:** avoid check-refs false-positive on pack name ([bd02ef0](https://github.com/event4u-app/agent-config/commit/bd02ef0b07d27653dbec4971b984d1d12b75ac5e))
+* **roadmap:** /video:from-song implementation roadmap + dashboard ([e4899c2](https://github.com/event4u-app/agent-config/commit/e4899c2c4f57b77bc62aa9187d4aa768ea6885f6))
+
+### Tests
+
+* **ai-video:** cover probe-audio segmentation + from-song registration ([9a58807](https://github.com/event4u-app/agent-config/commit/9a588071de1601337e6e65dc9b940b0a36dcfd42))
+
+### Chores
+
+* archive road-to-leaner-core-and-discovery (all phases complete) ([6b6b191](https://github.com/event4u-app/agent-config/commit/6b6b191f3b9ff328b9b7f3d6aa6eaf718e471934))
+* regenerate core pack manifest with skill/skill:preview commands ([9bd88c9](https://github.com/event4u-app/agent-config/commit/9bd88c9b7b18bdd16ef2c77066aa485d96fc2b2f))
+* register skills/skill/knowledge clusters + regen outputs ([1156566](https://github.com/event4u-app/agent-config/commit/1156566ad82647c596e47483d57b3fcb71f9a275))
+* bump command count 135 -> 136 in README badge + browse lines ([82c27c3](https://github.com/event4u-app/agent-config/commit/82c27c3383f53d29735784bae5fb86a151c19e39))
+* **ai-video:** regenerate manifests, counts, condensation hashes ([fe4dc28](https://github.com/event4u-app/agent-config/commit/fe4dc280437f209b31190a5117848feb92290015))
+
+Tests: 5336 (+65 since 5.4.1)
+
 ## [5.4.1](https://github.com/event4u-app/agent-config/compare/5.4.0...5.4.1) (2026-05-30)
 
 ### Documentation

package/CONTRIBUTING.md

@@ -237,6 +237,12 @@ Release notes live in [`CHANGELOG.md`](CHANGELOG.md) and are generated by
 the last tag. Contributors do **not** edit the changelog by hand; writing
 clean commit subjects is how notes are authored.
 
+The consumer-facing breaking changes per major are indexed for discovery in
+[`BREAKING_CHANGES.md`](BREAKING_CHANGES.md) — one row per `X.0.0`, each pointing
+back to the full `CHANGELOG.md` entry. That file is the discoverable index; this
+section and [`docs/contracts/CHANGELOG-conventions.md`](docs/contracts/CHANGELOG-conventions.md)
+remain the canonical policy.
+
 ### Runtime dependency floors — never pin to the freshest patch
 
 `dependencies` in `package.json` are resolved on the **consumer's** machine