Multi agent coordination
Skill Coinyak/onchainai/.agents/skills/multi-agent-coordination
Spawn and coordinate up to five OnchainAI subagents with exclusive path ownership, handoff packets, MCP observability rules, and verification tiers. Use when the task spans frontend + backend + schema, or when the user asks for parallel subagents, multi-agent work, or MCP-aware deploy/debug workflow.From its SKILL.md
npx -y skills add Coinyak/onchainai --skill multi-agent-coordinationAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
1.7 KB, as published. Nobody here has run it
Multi-Agent Coordination
Read before spawning subagents:
docs/MULTI_AGENT_COORDINATION.md— roster, DAG, verification matrixdocs/MCP_AGENT_WORKFLOW.md— Vercel/Railway/GitHub/onchainai routingdocs/handoff-packet-template.md— required between agents
Quick roster
| Role | Paths |
|---|---|
| Backend Core | src/ (API, crawler, MCP) |
| Frontend Surface | frontend/ |
| Data & Schema | migrations/, sqlx prepare |
| Harness & Deploy | scripts/, .github/, .grok/, .cursor/ |
| Security & Trust | auth, admin, install_safety, x402 |
Coordinator: DAG + merge handoffs + agent-harness-check.sh + user summary.
Spawn checklist
git status --short- Assign exclusive globs; no two writers on same file
- Serialize seams: schema → API contract packet → frontend
- MCP read-only by default; deploy only via
./scripts/deploy-railway.shwhen user asks - Handoff packet on every boundary
MCP (observe, not replace gates)
- vercel — Next deploy/logs (
onchain-ai/onchainai) - railway — API logs, env names, deploy status
- onchainai — live product queries
- github — PR/diff when user names a PR
Never paste secrets from MCP output. Never redeploy via MCP without explicit user approval.