ci-monitorlisted

# CI Monitor Monitor GitHub Actions CI health and respond to failures. Fixes simple issues automatically, escalates complex ones to the issue queue. ## Workflow ### Step 1: Check Main Branch CI ```bash gh run list --branch main --limit 5 --json status,conclusion,databaseId,headBranch,name,createdAt ``` Examine the results: - If all 5 runs have `conclusion: "success"` → **main is healthy**, proceed to Step 3 - If any run has `conclusion: "failure"` → proceed to Step 2 ### Step 2: Diagnose Failure For each failed run: ```bash gh run view <databaseId> --log-failed 2>&1 | head -200 ``` #### 2a: Check for Existing Fix ```bash gh issue list --label "ci-fix" --state open --search "<error keyword>" --json number,title ``` If an open `ci-fix` issue already exists for this failure, **skip it** — it is already being tracked. #### 2b: Classify the Failure **Simple failures** (auto-fixable): - Lint errors (`eslint`, `prettier`) - Type errors (`tsc`, missing imports, wrong types) - Missing Prisma client generation (`prisma generate`) - Test failures with obvious fixes (snapshot updates, assertion mismatches from recent changes) **Complex failures** (escalate to issue): - Flaky tests (passes on retry) - Infrastructure failures (Docker, database connectivity) - Multi-file refactoring needed - Dependency conflicts #### 2c: Auto-Fix Simple Failures ```bash mbe agent run "Fix CI failure: <concise description of error from logs>" --max-budget 0.50 --adapter auto ``` The agent