Agent Dashboard Sync

Operate dashboard data sync as a no-LLM pipeline.

Hard Rules

1. Keep high-frequency state out of Git commits.
2. Use Cloudflare Worker + KV for runtime sync.
3. Run collector from local cron (*/2 * * * *) and do not call LLM in collector path.
4. Never commit or print production tokens/secrets in files, logs, or screenshots.
5. Do not publish absolute paths in skill docs; use relative paths or placeholders (<PROJECTS_ROOT>, <SHARED_ROOT>).

Scope Boundary

• This skill owns: collector, Worker ingest/read API, KV schema, cron deployment, dashboard data source wiring.
• This skill does not own: cross-agent protocol, constitution governance, Discord routing rules.

KV Data Contract (v1)

• fleet:registry
• fleet:heartbeat:<agent_id>
• fleet:cron:<agent_id>
• fleet:runtime:<agent_id>
• fleet:events:recent
• fleet:updated_at

See references/schema.md for payload shape.

Minimal Rollout

1. Deploy Worker + KV namespace.
2. Configure dashboard env to cloudflare mode.
3. Install collector cron on each node with unique AGENT_ID.
4. Verify /health, then /fleet, then dashboard UI.

Security Checklist

• Store INGEST_TOKEN and READ_TOKEN as worker secrets.
• Keep dashboard read token server-side (DASHBOARD_READ_TOKEN), never client-exposed.
• Keep NEXT_PUBLIC_* vars non-sensitive only.
• Redact tokens before sharing commands/logs.

Runbook Links

• Worker setup and command sequence: references/worker-setup.md
• Collector and crontab setup: references/collector-cron.md
• Env variable matrix: references/env-matrix.md
• Data schema reference: references/schema.md
• Security policy for all agents: references/security-rules.md