davinci-resolve-mcp 2.135.0 → 2.136.0

package/AGENTS.md

@@ -66,8 +66,8 @@ semantics.
 - Public overview, current stats, and docs map: `README.md`
 - Historical release notes: `CHANGELOG.md`
 - AI assistant operating reference: `docs/SKILL.md`
-- Craft guidance for non-Claude-Code clients: the `knowledge` tool serves
-  `.claude/skills/`, `docs/guides/`, and `docs/kernels/` as resolved prose over
+- Craft guidance for remote clients: the `knowledge` tool serves
+  `.agents/skills/`, `docs/guides/`, and `docs/kernels/` as resolved prose over
   MCP. When you add a skill, guide, or kernel it is indexed automatically — a
   drift guard fails the suite if anything in those directories is unreachable
 - Release checklist and validation rules: `docs/process/release-process.md`
@@ -95,6 +95,12 @@ semantics.
 - Installer: `install.py`
 - Tests: `tests/`
 - Examples: `examples/`
+- Portable agent assets: `.agents/skills/`, `.agents/roles/`, `.agents/hooks/`
+- Client adapters: `.claude/` and `.codex/`. Skills are byte-identical in
+  `.claude/skills/` (edit either copy, then run
+  `scripts/agent-rules/sync_portable_assets.py`); hooks are canonical in
+  `.agents/hooks/` with thin shims; `.claude/agents/*` keep their frontmatter
+  (tools/model pins) over the shared `.agents/roles/*` bodies
 
 ## Common Commands
 
@@ -130,8 +136,10 @@ for Resolve scripting; 3.13/3.14 are accepted and verified on Resolve Studio
 Per-domain routers pair the live tools with their offline (advanced-server)
 counterparts. Every domain is available on demand in ANY MCP client as a slash
 prompt (`/color_grade_workflow`, `/timeline_edit_workflow`, `/conform_workflow`,
-`/delivery_workflow`, `/analyze_media`) and, in Claude Code, as a `.claude/skills/`
-skill. Full per-action depth lives in `docs/kernels/`.
+`/delivery_workflow`, `/analyze_media`) and as canonical skills in
+`.agents/skills/` (Claude Code loads content-complete `.claude/skills/` copies
+kept byte-identical by `scripts/agent-rules/sync_portable_assets.py` and its
+parity guard). Full per-action depth lives in `docs/kernels/`.
 
 - **Color / Grade** (`/color_grade_workflow`, skill `resolve-color`) — grading, correcting, shot matching, developing looks, or applying/modifying LUTs, CDLs, DRX grades, or copied grades. Live: timeline_item_color. Offline: drx. Depth: docs/kernels/color-grade-kernel.md + docs/guides/color-decision-guide.md.
 - **Timeline Edit** (`/timeline_edit_workflow`, skill `resolve-edit`) — cutting, trimming, pacing, duplicating/moving clips, copying ranges, building variants, tightening a cut, or generating an editorial changelist. Live: timeline. Offline: editorial. Depth: docs/kernels/timeline-edit-kernel.md + docs/guides/editorial-decision-guide.md.

package/docs/README.md

@@ -41,57 +41,63 @@ ignored scratch folders such as `docs/_scratch/`.
 - [Project Lifecycle](kernels/project-lifecycle-kernel.md)
 - [Extension Authoring](kernels/extension-authoring-kernel.md)
 
-## Claude Code Skills
-
-Per-domain skills in `.claude/skills/` route craft ↔ live tools ↔ offline
-advanced tools automatically when an agent works in that domain. They are thin
-bridges — the authoritative depth stays in the kernels and guides above.
-
-- `resolve-mcp` (`.claude/skills/resolve-mcp/SKILL.md`) — orientation/index: the map to the domain skills below (self-trigger; not an auto-loader)
-- `resolve-color` (`.claude/skills/resolve-color/SKILL.md`) — grading, looks, shot match, LUT/CDL/DRX
-- `resolve-edit` (`.claude/skills/resolve-edit/SKILL.md`) — cutting, ranges, variants, changelist
-- `resolve-conform` (`.claude/skills/resolve-conform/SKILL.md`) — conform, relink, finishing QC, grade tracing
-- `resolve-delivery` (`.claude/skills/resolve-delivery/SKILL.md`) — render, deliverable QC, media/provenance
-- `resolve-fusion` (`.claude/skills/resolve-fusion/SKILL.md`) — Fusion comps (titles, motion graphics, VFX)
-- `resolve-audio` (`.claude/skills/resolve-audio/SKILL.md`) — audio/Fairlight tracks, buses, loudness, sync
-- `resolve-media-pool` (`.claude/skills/resolve-media-pool/SKILL.md`) — media pool ingest, organize, multicam
-- `resolve-media-analysis` (`.claude/skills/resolve-media-analysis/SKILL.md`) — source-safe media intelligence
-
-Each skill is a directory containing `SKILL.md`. Claude Code does not discover
-loose `.md` files in `.claude/skills/`; a skill placed at the top level of that
-directory silently never loads.
+## Portable Agent Skills
+
+Per-domain skills in `.agents/skills/` route craft ↔ live tools ↔ offline
+advanced tools automatically when an agent works in that domain. This is the
+canonical, client-neutral source. Codex discovers it directly; Claude Code loads
+`.claude/skills/` copies that are CONTENT-COMPLETE and byte-identical to the
+canonical files — Claude routes on the frontmatter description at selection
+time, so a pointer stub would degrade skill routing. Edit either copy, then run
+`scripts/agent-rules/sync_portable_assets.py`; the parity guard
+(`tests/test_portable_asset_parity.py`) fails the suite on drift.
+
+- `resolve-mcp` — orientation/index: the map to the domain skills below (self-trigger; not an auto-loader)
+- `resolve-color` — grading, looks, shot match, LUT/CDL/DRX
+- `resolve-edit` — cutting, ranges, variants, changelist
+- `resolve-conform` — conform, relink, finishing QC, grade tracing
+- `resolve-delivery` — render, deliverable QC, media/provenance
+- `resolve-fusion` — Fusion comps (titles, motion graphics, VFX)
+- `resolve-audio` — audio/Fairlight tracks, buses, loudness, sync
+- `resolve-media-pool` — media pool ingest, organize, multicam
+- `resolve-media-analysis` — source-safe media intelligence
+
+Each canonical skill is a directory containing `SKILL.md`. The Claude adapter
+uses the same directory shape because Claude Code does not discover loose files.
 
 Two end-to-end assembly recipes sit alongside the domain skills. Where a domain
 skill routes, these two walk a whole job:
 
-- `resolve-rough-cut` (`.claude/skills/resolve-rough-cut/SKILL.md`) — **additive**:
+- `resolve-rough-cut` — **additive**:
   select shots from a folder of many clips into an assembled timeline.
-- `resolve-tighten-recording` (`.claude/skills/resolve-tighten-recording/SKILL.md`)
+- `resolve-tighten-recording`
   — **subtractive**: remove dead air from one long single-take recording.
 
 Three more sit outside the domain routing:
 
-- `house-style` (`.claude/skills/house-style/SKILL.md`) — accumulated editorial
-  corrections, so the same note is not given twice. Claude-only; append to it
+- `house-style` — accumulated editorial
+  corrections, so the same note is not given twice. Append to it
   when an editorial decision is corrected.
-- `resolve-session` (`.claude/skills/resolve-session/SKILL.md`) — `/resolve-session`
+- `resolve-session` — `/resolve-session`
   connects, confirms edition and bridge, and reports project/timeline/pool state.
-- `release-check` (`.claude/skills/release-check/SKILL.md`) — `/release-check`
+- `release-check` — `/release-check`
   walks a version bump using [docs/process/release-process.md](process/release-process.md)
   as the sole source; the skill wraps that doc, it does not duplicate it.
 
 The offline half of every one is the advanced server; see
 [Advanced Server](../resolve-advanced/README.md).
 
-## Claude Code Hooks and Subagents
+## Portable Hooks and Reviewers
 
-Four hooks live in `.claude/hooks/` — two `PreToolUse` guards enforcing rules
-`AGENTS.md` states in prose, and two `PostToolUse` checks that surface
-engineering drift right after the edit that caused it instead of at the next
-test run. All four ship as scripts but are **not** wired up by default — the
-repository does not enable hooks on your behalf. Opt in by adding the block
-below to your own `.claude/settings.local.json` (gitignored, so it stays
-yours):
+Four canonical hooks live in `.agents/hooks/` — two `PreToolUse` guards
+enforcing rules `AGENTS.md` states in prose, and two `PostToolUse` checks that
+surface engineering drift immediately after the edit that caused it. Thin
+wrappers in `.claude/hooks/` and `.codex/hooks/` translate only the host
+protocol; policy stays in the canonical scripts.
+
+Codex wiring is committed in `.codex/hooks.json`. Because project hooks execute
+commands, review and trust that file through Codex's `/hooks` UI. Claude Code
+users opt in by adding this block to `.claude/settings.local.json`:
 
 ```json
 {
@@ -139,18 +145,18 @@ yours):
 }
 ```
 
-Hooks are read at session start, so restart Claude Code after adding them. The
-four are:
+Hooks are read at session start, so restart the client after enabling them. The
+four canonical policies are:
 
-- `.claude/hooks/frame_verification_guard.py` — denies grade-applying actions on
+- `.agents/hooks/frame_verification_guard.py` — denies grade-applying actions on
   `timeline_item_color` until the session has actually looked at a
   Resolve-rendered frame, and asks before whole-grade artifacts
   (`safe_copy_grade`, `bulk_match_to_hero`) overwrite hand-work. `dry_run`
   passes through untouched.
-- `.claude/hooks/source_media_guard.py` — denies shell commands that write,
+- `.agents/hooks/source_media_guard.py` — denies shell commands that write,
   move, or delete source media outside a scratch root. Reads (`ffprobe`, and
   `ffmpeg` writing into scratch) pass.
-- `.claude/hooks/agent_rules_drift_check.py` — after an edit to anything
+- `.agents/hooks/agent_rules_drift_check.py` — after an edit to anything
   `generate.mjs` reads (`docs/SKILL.md`, `docs/kernels/README.md`,
   `resolve-advanced/README.md`, and `generate.mjs` itself, which carries the
   DOMAINS manifest inline) or to `AGENTS.md`, which it writes, runs
@@ -160,7 +166,7 @@ four are:
   blocks. Skips quietly if `node` isn't on `PATH`. **If `generate.mjs` grows a
   new input, add it to `SOURCE_PATHS` in the hook** — an unwatched input is a
   silent hook on exactly the edit it exists to catch.
-- `.claude/hooks/run_matching_test.py` — after an edit to `src/<module>.py`,
+- `.agents/hooks/run_matching_test.py` — after an edit to `src/<module>.py`,
   runs the matching `tests/test_<module>.py` if one exists, using the project
   venv (`venv/bin/python`) so `pytest` is actually importable. Informational
   only; skips quietly if there's no matching test file or no working `pytest`.
@@ -169,8 +175,13 @@ four are:
   for `src/server.py` or `src/granular/common.py`. Silence means "no matching
   test file", not "this edit is fine".
 
-Three review subagents in `.claude/agents/` run in their own context so bulky
-output (frame images, full test transcripts) stays out of the main session:
+Codex does not currently support an interactive `ask` result from `PreToolUse`.
+For the two whole-grade actions, its adapter adds a warning while the MCP
+server's confirmation-token gate remains authoritative. Hard denials, including
+missing frame evidence and source-media writes, block in both clients.
+
+Three canonical reviewer roles live in `.agents/roles/`, with native Codex
+definitions under `.codex/agents/` and Claude adapters under `.claude/agents/`:
 
 - `cut-reviewer` — screens an assembled timeline from its frames and reports on
   pacing, shot order, continuity, and coverage gaps.

package/docs/SKILL.md

@@ -207,7 +207,7 @@ Operating rules an agent must know:
 
 Per-domain depth for both servers lives in the kernels (`docs/kernels/`), each of
 which carries an "Advanced (offline) server" section where an offline counterpart
-exists. Claude Code skills in `.claude/skills/` (`resolve-color`, `resolve-edit`,
+exists. Portable skills in `.agents/skills/` (`resolve-color`, `resolve-edit`,
 `resolve-conform`, `resolve-delivery`, `resolve-media-analysis`) route
 craft ↔ live ↔ offline automatically when working in that domain.
 

package/docs/kernels/README.md

@@ -38,7 +38,7 @@ two:
 - Timeline Edit → `editorial` (interchange + changelist), `drt`
 
 Per-domain Claude Code skills route craft ↔ live ↔ offline automatically and live
-in `.claude/skills/`: `resolve-color`, `resolve-edit`, `resolve-conform`,
+in `.agents/skills/`: `resolve-color`, `resolve-edit`, `resolve-conform`,
 `resolve-delivery` (plus `resolve-media-analysis`). `docs/SKILL.md` is the
 top-level brief for both servers.
 

package/docs/kernels/audio-fairlight-kernel.md

@@ -121,7 +121,7 @@ patches audio **offline, no Resolve running**:
 
 Rule of thumb: plan/measure offline, apply mix/track changes live; use
 `fairlight` for bus work the scripting API can't reach. See the `resolve-audio`
-skill (`.claude/skills/resolve-audio/SKILL.md`) and the `/audio_workflow` prompt.
+skill (`.agents/skills/resolve-audio/SKILL.md`) and the `/audio_workflow` prompt.
 
 ## Live Probe
 

package/docs/kernels/color-grade-kernel.md

@@ -156,7 +156,7 @@ Cross-server rules an agent must know:
 - **Deps.** The grading catalog needs `sharp`; call the advanced `capabilities`
   tool for live status and install hints.
 
-See the `resolve-color` skill (`.claude/skills/resolve-color/SKILL.md`) for the
+See the `resolve-color` skill (`.agents/skills/resolve-color/SKILL.md`) for the
 craft ↔ live ↔ offline routing and the frame-first rule.
 
 ## Live Probe

package/docs/kernels/fusion-composition-kernel.md

@@ -119,7 +119,7 @@ running**, via the `fusion` tool:
 Rule of thumb: author and verify the comp offline, then apply it live with
 `fusion_comp` `safe_add_tool` → `safe_set_inputs` → `safe_connect_tools` (the
 `to_api_calls` output maps directly onto those). See the `resolve-fusion` skill
-(`.claude/skills/resolve-fusion/SKILL.md`) and the `/fusion_workflow` prompt.
+(`.agents/skills/resolve-fusion/SKILL.md`) and the `/fusion_workflow` prompt.
 
 ## Live Probe
 

package/docs/kernels/media-pool-ingest-kernel.md

@@ -137,7 +137,7 @@ actions):
 Rule of thumb: verify and inventory the card offline *before* importing, then
 import/organize live. `media` also serves the delivery side (see the
 `resolve-delivery` skill). See the `resolve-media-pool` skill
-(`.claude/skills/resolve-media-pool/SKILL.md`) and the `/media_pool_workflow` prompt. Never
+(`.agents/skills/resolve-media-pool/SKILL.md`) and the `/media_pool_workflow` prompt. Never
 rename or derive camera originals without explicit approval.
 
 ## Live Evidence

package/docs/kernels/render-deliver-kernel.md

@@ -135,7 +135,7 @@ Rules an agent must know:
 - **Deps.** `deliverable`/`media` QC needs **ffmpeg + ffprobe on PATH** (GPL, not
   bundled) — call the advanced `capabilities` tool for status + install hints.
 
-See the `resolve-delivery` skill (`.claude/skills/resolve-delivery/SKILL.md`) for the
+See the `resolve-delivery` skill (`.agents/skills/resolve-delivery/SKILL.md`) for the
 craft ↔ live ↔ offline routing.
 
 ## Live Evidence

package/docs/kernels/timeline-conform-interchange-kernel.md

@@ -160,7 +160,7 @@ Gotchas the live path shares:
 - **Deps.** `better-sqlite3` gates lineage/reverse/DB; `sharp`/ffmpeg gate frame
   compare — call the advanced `capabilities` tool.
 
-See the `resolve-conform` skill (`.claude/skills/resolve-conform/SKILL.md`) for the
+See the `resolve-conform` skill (`.agents/skills/resolve-conform/SKILL.md`) for the
 craft ↔ live ↔ offline routing.
 
 ## Live Probe

package/docs/kernels/timeline-edit-kernel.md

@@ -196,7 +196,7 @@ Use these to answer "what changed between v3 and v4" or to hand a conform an
 accurate change list without opening either timeline. For conforming/relinking
 that change list, see the Timeline Conform / Interchange kernel and the
 `resolve-conform` skill; for the edit ↔ offline routing, see the `resolve-edit`
-skill (`.claude/skills/resolve-edit/SKILL.md`).
+skill (`.agents/skills/resolve-edit/SKILL.md`).
 
 ## Development Guardrails
 

package/install.py

@@ -37,7 +37,7 @@ from src.utils.update_check import (
 
 # ─── Version ──────────────────────────────────────────────────────────────────
 
-VERSION = "2.135.0"
+VERSION = "2.136.0"
 # Only hard floor: mcp[cli] requires Python 3.10+. There is no upper bound —
 # Resolve's scripting bridge loads into newer interpreters on recent builds
 # (Python 3.14 verified against Resolve Studio 20.3.2). Older Resolve builds

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "davinci-resolve-mcp",
-  "version": "2.135.0",
+  "version": "2.136.0",
   "description": "NPM bootstrapper for the DaVinci Resolve MCP Server.",
   "license": "MIT",
   "author": "Samuel Gursky <samgursky@gmail.com>",

package/scripts/agent-rules/README.md

@@ -24,7 +24,8 @@ cannot drift.
    | Platform | File(s) | Trigger | Source |
    |---|---|---|---|
    | Codex / OpenCode / Zed / others | `AGENTS.md` (`## Domain Routing` block) | always-on | generated block |
-   | Claude Code | `.claude/skills/<name>/SKILL.md` | semantic (description) | hand-authored (rich) |
+   | Codex and compatible agents | `.agents/skills/<name>/SKILL.md` | semantic (description) | hand-authored canonical skill |
+   | Claude Code | `.claude/skills/<name>/SKILL.md` | semantic (description) | thin adapter to canonical skill |
    | Cursor | `.cursor/rules/*.mdc` | `alwaysApply` repo rule + per-domain `description` (agent-requested) | generated |
    | VS Code / Copilot | `.github/copilot-instructions.md` + `.github/instructions/*.instructions.md` | always-on + `applyTo` | generated |
    | Windsurf | `.windsurf/rules/*.md` + `.windsurfrules` | rules dir + legacy flat | generated |
@@ -33,8 +34,9 @@ cannot drift.
    | Continue | `.continue/rules/resolve-mcp.md` | always-on | generated |
    | Claude Desktop | — (chat client, no repo rules) | — | MCP prompts only |
 
-   `.claude/skills/*` stays hand-authored (Claude's rich semantic skills). Each
-   one is a **directory** holding a `SKILL.md` named for its frontmatter `name`;
+   `.agents/skills/*` stays hand-authored as the canonical rich skill source.
+   `.claude/skills/*` contains only compatibility adapters. Each skill is a
+   **directory** holding a `SKILL.md` named for its frontmatter `name`;
    a loose `.md` at the top level of `.claude/skills/` is never discovered and
    fails silently. Every
    other file is generated. `AGENTS.md` remains the universal backstop for any
@@ -65,7 +67,7 @@ node scripts/agent-rules/generate.mjs --check   # exit 1 if anything is stale (C
 
 - **New domain:** add an entry to `DOMAINS` in `generate.mjs`, add a matching
   `@mcp.prompt` in `src/server.py`, and (optionally) a rich
-  `.claude/skills/<name>/SKILL.md`. Regenerate.
+  `.agents/skills/<name>/SKILL.md` and add/update its client adapter. Regenerate.
 - **New platform:** add an `emit(...)` for its convention in `generate.mjs`,
   reusing `domainBody(d)` / `repoHygiene`. Regenerate. Do not hand-edit generated
   files — the drift guard will fail.

package/scripts/agent-rules/sync_portable_assets.py

@@ -0,0 +1,109 @@
+#!/usr/bin/env python3
+"""Sync the portable agent assets (`.agents/`) with their Claude adapters.
+
+The layout contract (adapted from PR #173):
+
+- `.agents/skills/*/SKILL.md` is the canonical, host-neutral skill corpus —
+  the `knowledge` MCP tool serves it, and non-Claude hosts read it directly.
+- `.claude/skills/*/SKILL.md` is a CONTENT-COMPLETE copy, byte-identical to
+  the canonical file. Claude Code routes on the frontmatter description at
+  selection time, so the adapter must carry the full text — a pointer stub
+  would cost a read per invocation and hide the trigger phrasing.
+- `.agents/roles/*.md` is the body (frontmatter stripped) of the matching
+  `.claude/agents/*.md`. The Claude file keeps its frontmatter — name,
+  description, tools, and the deliberate `model:` pin — which host-neutral
+  role files cannot carry.
+- `.agents/hooks/*.py` is canonical executable logic; `.claude/hooks` and
+  `.codex/hooks` are thin runpy shims (execution-only surfaces need no
+  content duplication).
+
+Edit EITHER side of a skill (they are identical); edit `.claude/agents` for
+roles. Then run this script to restore the invariant:
+
+    venv/bin/python scripts/agent-rules/sync_portable_assets.py          # write
+    venv/bin/python scripts/agent-rules/sync_portable_assets.py --check  # verify only
+
+`tests/test_portable_asset_parity.py` fails the suite when the invariant is
+broken, so drift cannot land.
+"""
+
+from __future__ import annotations
+
+import argparse
+import re
+import sys
+from pathlib import Path
+
+REPO = Path(__file__).resolve().parents[2]
+CLAUDE_SKILLS = REPO / ".claude" / "skills"
+AGENT_SKILLS = REPO / ".agents" / "skills"
+CLAUDE_AGENTS = REPO / ".claude" / "agents"
+AGENT_ROLES = REPO / ".agents" / "roles"
+
+_FRONTMATTER_RE = re.compile(r"^---\n.*?\n---\n\n?", re.S)
+
+
+def strip_frontmatter(text: str) -> str:
+    m = _FRONTMATTER_RE.match(text)
+    return text[m.end():] if m else text
+
+
+def skill_pairs():
+    names = sorted(
+        {p.parent.name for p in CLAUDE_SKILLS.glob("*/SKILL.md")}
+        | {p.parent.name for p in AGENT_SKILLS.glob("*/SKILL.md")}
+    )
+    for name in names:
+        yield name, CLAUDE_SKILLS / name / "SKILL.md", AGENT_SKILLS / name / "SKILL.md"
+
+
+def role_pairs():
+    names = sorted(
+        {p.stem for p in CLAUDE_AGENTS.glob("*.md")} | {p.stem for p in AGENT_ROLES.glob("*.md")}
+    )
+    for name in names:
+        yield name, CLAUDE_AGENTS / f"{name}.md", AGENT_ROLES / f"{name}.md"
+
+
+def run(check_only: bool) -> int:
+    problems: list[str] = []
+    for name, claude_p, agent_p in skill_pairs():
+        if not claude_p.exists():
+            problems.append(f"skill '{name}': canonical exists but .claude adapter is missing — it would never load in Claude Code")
+            if not check_only and agent_p.exists():
+                claude_p.parent.mkdir(parents=True, exist_ok=True)
+                claude_p.write_text(agent_p.read_text())
+            continue
+        src = claude_p.read_text()
+        if not agent_p.exists() or agent_p.read_text() != src:
+            problems.append(f"skill '{name}': .agents copy {'missing' if not agent_p.exists() else 'differs'}")
+            if not check_only:
+                agent_p.parent.mkdir(parents=True, exist_ok=True)
+                agent_p.write_text(src)
+    for name, claude_p, role_p in role_pairs():
+        if not claude_p.exists():
+            problems.append(f"role '{name}': .agents/roles exists but .claude/agents is missing — restore the frontmatter file by hand (it carries tools/model pins this script cannot invent)")
+            continue
+        body = strip_frontmatter(claude_p.read_text())
+        if not role_p.exists() or role_p.read_text() != body:
+            problems.append(f"role '{name}': .agents/roles {'missing' if not role_p.exists() else 'differs from the .claude/agents body'}")
+            if not check_only:
+                role_p.parent.mkdir(parents=True, exist_ok=True)
+                role_p.write_text(body)
+    if problems:
+        for p in problems:
+            print(("DRIFT: " if check_only else "SYNCED: ") + p)
+        return 1 if check_only else 0
+    print("portable assets in sync")
+    return 0
+
+
+def main() -> int:
+    ap = argparse.ArgumentParser(description=__doc__)
+    ap.add_argument("--check", action="store_true", help="verify only; non-zero exit on drift")
+    args = ap.parse_args()
+    return run(check_only=args.check)
+
+
+if __name__ == "__main__":
+    sys.exit(main())

package/src/granular/common.py

@@ -87,7 +87,7 @@ if not logging.getLogger().handlers:
         handlers=[logging.StreamHandler()],
     )
 
-VERSION = "2.135.0"
+VERSION = "2.136.0"
 logger = logging.getLogger("davinci-resolve-mcp")
 logger.info(f"Starting DaVinci Resolve MCP Server v{VERSION}")
 logger.info(f"Detected platform: {get_platform()}")

package/src/server.py

@@ -11,7 +11,7 @@ Usage:
     python src/server.py --full       # Start the 353-tool granular server instead
 """
 
-VERSION = "2.135.0"
+VERSION = "2.136.0"
 
 import base64
 import os
@@ -545,8 +545,8 @@ def prep_color_handoff(output_dir: str = "") -> str:
 # ─── Per-domain workflow routers ───────────────────────────────────────────────
 # Cross-platform depth: these surface as slash commands in EVERY MCP client
 # (Codex, Cursor, Copilot, Continue, Claude Desktop, …), so per-domain routing is
-# not limited to Claude Code's .claude/skills/. They mirror the repo skills of the
-# same name; keep them in sync with docs/kernels/*.
+# not limited to one client's adapters. They mirror the canonical `.agents/skills/`
+# skills of the same name; keep them in sync with docs/kernels/*.
 
 
 @mcp.prompt(

package/src/utils/knowledge.py

@@ -2,7 +2,7 @@
 
 This repository carries a real body of craft guidance: how to tighten a recording
 without cutting the breath out of it, what to look at before applying a grade, which
-Resolve API calls silently lie. It lives in two places — `.claude/skills/*/SKILL.md`
+Resolve API calls silently lie. It lives in two places — `.agents/skills/*/SKILL.md`
 and `docs/guides` + `docs/kernels` — and until now both were reachable only by an
 agent that could open files in this checkout.
 
@@ -57,7 +57,7 @@ from pathlib import Path
 from typing import Any, Dict, List, Optional, Tuple
 
 REPO_ROOT = Path(__file__).resolve().parents[2]
-SKILLS_DIR = REPO_ROOT / ".claude" / "skills"
+SKILLS_DIR = REPO_ROOT / ".agents" / "skills"
 GUIDES_DIR = REPO_ROOT / "docs" / "guides"
 KERNELS_DIR = REPO_ROOT / "docs" / "kernels"
 
@@ -217,24 +217,10 @@ def _first_paragraph(body: str) -> str:
 def _title_of(body: str, fallback: str) -> str:
     for line in body.splitlines():
         if line.startswith("# "):
-            # `# Resolve Tighten Recording — Claude Code Skill` is scaffolding for one
-            # client. The subject is everything before the dash.
-            title = line[2:].strip()
-            return re.sub(r"\s*[—-]\s*Claude Code Skill\s*$", "", title).strip()
+            return line[2:].strip()
     return fallback
 
 
-def _strip_client_scaffolding(body: str) -> str:
-    """Drop the one H1 that names a specific client. Prose is left alone."""
-    return re.sub(
-        r"^#\s+(.*?)\s*[—-]\s*Claude Code Skill\s*$",
-        lambda m: f"# {m.group(1)}",
-        body,
-        count=1,
-        flags=re.M,
-    )
-
-
 _DOC_REF_RE = re.compile(r"`(docs/[^`\s]+\.md)`|\]\((docs/[^)\s]+\.md)\)")
 
 
@@ -275,7 +261,7 @@ def _build_index() -> Dict[str, Dict[str, Any]]:
                 "summary": fields.get("description", "") or _first_paragraph(body),
                 "category": "repo" if topic in _REPO_SKILLS else "workflow",
                 "path": str(skill_md.relative_to(REPO_ROOT)),
-                "body": _strip_client_scaffolding(body).strip(),
+                "body": body.strip(),
             }
 
     for category, directory in (("guide", GUIDES_DIR), ("kernel", KERNELS_DIR)):