ableton-mind-build
Skill9 starsChanged 5 months agoMITView on GitHub
Orchestrates the 8-agent ableton-mind team for concrete build, scaffold, phase execution, local-copilot work, continuation, QA gate or roadmap-cycle work. For broad quality audits, run-all-commands sweeps, security/usability/refactor review or coverage hardening, use ableton-mind-quality-audit instead.
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 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.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Use this skill whenever the user asks to execute, implement, continue, scaffold, redo, synchronize, evaluate or gate concrete work for …
- Very specific: line 28
Read _workspace/PROGRESS.md if it exists.
- Distinctive: line 51
When work returns, read each _workspace/*_summary.md and the QA report.
- Distinctive: line 58
Never run two workers against the same ownership boundary in the same cycle.
- Distinctive: line 60
Keep English as the default repo language.
≈ “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
ableton-mind-build — Team Orchestrator
Use this skill whenever the user asks to execute, implement, continue, scaffold, redo, synchronize, evaluate or gate concrete work for ableton-mind. Conceptual questions about PLAN.md can be answered directly without launching a full cycle.
For broad repo quality audits, command sweeps, security/usability/refactor review, missing-test coverage or release-readiness hardening, hand off to ableton-mind-quality-audit. That sibling harness uses a different team and report structure while preserving this build harness for PLAN.md phase execution.
Team
| Track | Agent | Domain |
| Lead | architect | plan, ADRs, integration, phase gates |
| A — Server | ts-server-engineer | TS/Node MCP server, tools, resources, prompts |
| A — Bridge | python-bridge-engineer | Python Remote Script, LiveAPI, TCP JSON-RPC |
| B — Knowledge | knowledge-curator | device schemas, packs, scales, MIDI data |
| C — Recipes | recipe-designer | JSON music recipes |
| D — Distribution | distribution-docs-engineer | MCPB/DXT, npm, Docker, docs, CI, CLI |
| E — QA | qa-integration | parity, contract drift, recipe lint, smoke, gate |
| F — Local Copilot | local-copilot-engineer | local LLM, chat, ask, tool tiers, safety policy |
Phase 0 — Always Gather Context First
- Read PLAN.md.
- Read _workspace/PROGRESS.md if it exists.
- List _workspace/decisions/ and _workspace/qa/.
- List _workspace/cycle-briefing-*.md.
- Choose the mode:
| Mode | When | Action |
| Initial | PROGRESS.md does not exist | Create workspace state and plan Phase 0 spike |
| Continue | PROGRESS.md points to ongoing work | Write the next cycle briefing and continue |
| Redo phase | User requests a redo | Archive current workspace and recreate the requested scope |
| Phase gate | QA exists but PROGRESS is stale | Read verdict, update PROGRESS and decide next cycle |
| Sync only | User asks status | Report state without dispatching tracks |
Cycle Briefing
The architect writes _workspace/cycle-briefing-{N}.md with: PLAN phase, one-sentence goal, assignments per track, changed contracts, dependencies and gate criteria. Keep tasks concrete and file-scoped.
Execution
Run tracks in parallel only when their write scopes do not conflict and the current tooling/user permission allows sub-agents. Phase 0 spike is sequential: contract, bridge, server, integration. Later phases can run server, bridge, knowledge, recipes, distribution, local-copilot and QA in parallel as dependencies allow. Local-copilot work owns src/llm/ and src/cli/chat.ts; coordinate with distribution for public docs and with QA for safety-tier tests.
Integration
When work returns, read each _workspace/*_summary.md and the QA report. Compare delivered work with the briefing, write or update ADRs for contract changes, update PROGRESS.md and report the outcome in under 200 words.
Invariants
- Nothing lands in production paths without QA evidence.
- Contract mutations require an ADR.
- QA is a gate input; the architect makes the final phase decision.
- Never run two workers against the same ownership boundary in the same cycle.
- Preserve unrelated local changes.
- Keep English as the default repo language; localized pages belong under docs/pt only.
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-quality-auditSkill
- ableton-remote-scriptSkill
- command-surface-auditSkill
- cross-boundary-qaSkill
- device-schema-extractionSkill
- local-copilotSkill
- 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.
Reviews can't be read right now.
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.