Integration Errors Until Clean

/loop 30m Start the "Integration Errors Until Clean" loop.

Inspired by Zapier (https://zapier.com).

Goal: all integration runs green for 24h or failures owned with remediation plan
Max iterations: 24
Between iterations run: Report failing zaps, error rate by integration, stale auth tokens, and MCP action failures
Exit when: 24h clean run history with no unowned failures and auth tokens refreshed

## Before you start
- Connect Slack (required) — Failure alerts
- Connect GitHub (optional) — Store runbooks and fixes
- Attach Loop runner (required) — Self-pace iterations and run checks between passes.
- Attach Test repair (required) — Run tests and fix failures without weakening coverage.
- Attach Code change + verification (required) — Edit code or configs and verify locally.
- Zapier (read via browser) — Workflow monitor
- Connected SaaS apps (read via api) — Trigger and action endpoints
- MCP (read via mcp) — Agent-triggered actions

## Steps
1. Monitor runs: Ingest failure logs and error rates across zaps and agent chains. [tools: Zapier, Slack]
2. Diagnose root cause: Classify auth, schema drift, rate limit, or payload errors. [tools: Connected SaaS apps]
3. Patch workflow: Update mappings, refresh OAuth, or adjust retry policy. [tools: Zapier, GitHub]
4. Validate fix: Replay failed payloads in test path. [tools: Zapier, MCP]
   Command: npm run integrations:smoke
5. Confirm clean window: Verify 24h green history before closing. [tools: Zapier]

Self-pace this loop. After each iteration, run the check command, read the output, and only continue if the exit condition is not met. Stop when the exit condition passes or max iterations is reached. Give a short status update each pass.

Guardrails:
- Do not disable error handling to force green
- Rotate secrets via approved store only
- Test fixes in duplicate zap before prod cutover