local-copilot
Skill9 starsChanged 5 months agoMITView on GitHub
Use this for any ableton-mind local LLM/copilot work: Ollama, OpenAI-compatible endpoints, `ableton-mind chat`, `ableton-mind ask`, `llm-run`, tool tiers, safe/default read-only policy, local browser UI, headless prompts, model pull, handoff, or docs/tests for the local copilot. Also use for follow-ups like re-run, update, debug, harden, add a model backend, or port more tdmcp copilot behavior.
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 3 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Use this skill when adding, debugging or extending the ableton-mind local LLM feature.
- Very specific: line 12
Read src/llm/config.ts, src/llm/tools.ts, src/llm/agent.ts, src/llm/server.ts and src/cli/chat.ts.
- Very specific: line 13
Read tests/llm-local-copilot.test.ts and the latest relevant QA report.
- Very specific: line 14
Check _workspace/tdmcp-compatible-features.md for deferred tdmcp-inspired items.
- Very specific: line 19
Keep the default tier read-only (safe).
- Distinctive: line 20
Never expose destructive or broad routing/mixer tools to a local model without a safety review.
- Very specific: line 23
Do not copy TouchDesigner-specific prompts, paths or tools from tdmcp.
- Very specific: line 27
Prefer explicit JSON Schemas in src/llm/tools.ts over new schema-conversion dependencies.
- Distinctive: line 29
Add or update tests whenever changing config defaults, tool tier membership, agent loop behavior, CLI parsing or server endpoints.
- Distinctive: line 34
Run focused checks first:
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
local-copilot
Use this skill when adding, debugging or extending the ableton-mind local LLM feature.
Context Check
- Read
src/llm/config.ts,src/llm/tools.ts,src/llm/agent.ts,src/llm/server.tsandsrc/cli/chat.ts. - Read
tests/llm-local-copilot.test.tsand the latest relevant QA report. - Check
_workspace/tdmcp-compatible-features.mdfor deferred tdmcp-inspired items. - Confirm whether the request is UI chat, headless
ask, model/backend, safety tier, docs, or QA.
Safety Rules
- Keep the default tier read-only (
safe). Use explicit--writefor simple mutations and--creativefor recipes/browser load. - Never expose destructive or broad routing/mixer tools to a local model without a safety review.
- Preserve the bridge-offline behavior: the chat UI can open without Live; live tools must return clear bridge errors.
- Browser server stays loopback-only and must reject non-loopback Host/Origin.
- Do not copy TouchDesigner-specific prompts, paths or tools from
tdmcp.
Implementation Rules
- Prefer explicit JSON Schemas in
src/llm/tools.tsover new schema-conversion dependencies. - The local copilot should use existing
ToolDefinitionhandlers, not duplicate Ableton business logic. - Add or update tests whenever changing config defaults, tool tier membership, agent loop behavior, CLI parsing or server endpoints.
- Public docs belong in English root docs and matching Portuguese pages under
docs/pt/.
Verification
Run focused checks first:
npm run typecheck
npx vitest run tests/llm-local-copilot.test.tsFor broader changes, add:
npm run lint
npm run test
npm run buildLive/Ollama validation is optional unless the task changes runtime connectivity. If skipped, report it explicitly.
Test Scenarios
- Normal:
ableton-mind ask --read-only "What is in this set?"with Ollama reachable and bridge connected. - Offline bridge:
ableton-mind chat --no-ollama --no-openopens the local UI even when Live is closed; live tool calls return a bridge-offline error. - Safety: default config resolves to
safe;--writeswitches tostandard;--creativeexposes recipes/browser load.
In this repository23 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- ableton-mind-buildSkill
- ableton-remote-scriptSkill
- cross-boundary-qaSkill
- device-schema-extractionSkill
- mcp-distributionSkill
- mcp-server-developmentSkill
- recipe-authoringSkill
- ableton-mind-buildSkill
- ableton-mind-quality-auditSkill
- ableton-remote-scriptSkill
- command-surface-auditSkill
- cross-boundary-qaSkill
- device-schema-extractionSkill
- mcp-distributionSkill
- mcp-server-developmentSkill
- recipe-authoringSkill
- refactor-maintainabilitySkill
- runtime-release-auditSkill
- security-auditSkill
- test-coverage-expansionSkill
- usability-flow-auditSkill
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

