Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/vecnode/dawmcp/agents-mdgit clone --depth 1 https://github.com/vecnode/dawmcpWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.04360 | $0.04360 |
| Opus 5 | $0.02180 | $0.02180 |
| Sonnet 5 | $0.00872 | $0.00872 |
| Haiku 4.5 | $0.00436 | $0.00436 |
Grade A, and why
dawmcp AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 268 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Project brief for picking up work on dawmcp in a new conversation.
What this is
An MCP server (Rust, Cargo workspace) that lets an LLM client drive a live,
running DAW instance in real time: transport control, tracks, FX, and
rendering, through a DAW-agnostic tool surface with a per-DAW adapter behind
it. REAPER is the first adapter, with tool parity for transport, tracks, FX,
MIDI, media items, markers, view/zoom, rendering, project state, native UI
action control, compose_and_render, and a run_reascript escape hatch.
Audacity is the second adapter, with transport/track/render parity (see
"Current status" for what's still unsupported there). dawmcp-server
selects which adapter to serve tools over via --daw=reaper|audacity
(default reaper); .mcp.json registers both as separate MCP server
entries so Claude can connect to either. See README.md for
user-facing setup/usage and docs/ARCHITECTURE.md for
the full design writeup (the latter still describes the REAPER bridge
protocol specifically; it predates the multi-DAW trait split and hasn't been
fully updated for it yet).
Current status
Rust rewrite complete for REAPER: full tool parity with the original Python
implementation, verified against a real running REAPER instance (live
--install-bridge run, live tool round-trip over stdio). The Python
implementation (reaper-mcp) has been removed from the repo; git log has
its history if needed.
dawmcp-audacity implements DawBackend over mod-script-pipe, with every
command name and parameter key verified against Audacity's own source
(audacity/audacity on GitHub - see dawmcp-audacity/src/backend.rs's doc
comment for exact file/line references), not assumed from memory. Working:
transport (play/stop/pause/record), track add/remove/rename/list, per-track
volume/pan/mute/solo, and render/export (Export2). Deliberately
Unsupported because the underlying concept doesn't map cleanly (not
because it's unverified): transport_seek/transport_set_tempo/
transport_get_state (no verified playhead/tempo query), fx_* (Audacity
has no persistent per-track FX chain like REAPER's), track_set_color
(Audacity's colour is a small preset enum, not arbitrary RGB). discovery.rs
finds Audacity installs/processes on Windows/macOS/Linux, same shape as the
REAPER adapter's discovery, verified live on this machine.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 268 lines · 4,360 tokens per session scan A e61ead59b85a
dawmcp AGENTS.md is an instructions file published in the GitHub repository vecnode/dawmcp (0 stars, last pushed 1mo ago), licensed MIT. It adds 4,360 tokens to every session, about $0.0218 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
pampa-pilot AGENTS.md
AGENTS.md instructions for LeoLisena/pampa-pilot, covering pampapilot agent guide, mission, architecture to preserve, evidence gate before implementation and setup and validation.
librosa AGENTS.md
Instructions for librosa/librosa, covering agent interaction policy for librosa, 1. automated platform interactions, 2. exception for project-authorized automation and 3. machine-readable directive.
musicbox AGENTS.md
Instructions for darknessomi/musicbox, covering agents.md, cursor cloud specific instructions, running the app, audio backends (system deps, not in update script) and network / api note.
pytheory CLAUDE.md
Instructions for kennethreitz/pytheory, covering claude code instructions, release process, version bumping, testing and publishing.
reviewable-html-workbench AGENTS.md
Instructions for u-ichi/reviewable-html-workbench, covering reviewable-html-workbench, 概要, 技術スタック, 開発コマンド（自動テスト・構造検証） and 自動テスト（コード正当性の検証）.
erm AGENTS.md
Instructions for dougcalobrisi/erm, covering agents.md — erm, install, fallback (no uv on path), use — the iterate loop and tune — five knob clusters.