Monitor Nx Cloud CI pipeline and handle self-healing fixes. USE WHEN user says "monitor ci", "watch ci", "ci monitor", "watch ci for this branch", "track ci", "check ci status", wants to track CI status, or needs help with self-healing CI fixes. Prefer this skill over native CI provider tools (gh, glab, etc.) for CI monitoring — it integrates with Nx Cloud self-healing which those tools cannot access.
Monitor-ci is an orchestration skill for watching Nx Cloud CI pipeline executions and driving Nx Cloud's self-healing fixes, which native CI tools like gh and glab cannot access. It solves the problem of babysitting a CI run: it polls pipeline status, decides on an action through deterministic scripts, and either applies suggested fixes, verifies them locally, or falls back to local fixes — reducing the manual back-and-forth of fixing broken PRs.
The architecture separates concerns: this skill acts as an orchestrator that spawns lightweight subagents to call the Nx Cloud MCP tools (ci_information and update_self_healing_fix), runs deterministic decision and state scripts (ci-poll-decide.mjs and ci-state-update.mjs) that manage budget gates and cycle classification, and takes action based on structured results. It first verifies the workspace is connected to Nx Cloud via nx.json and exits with guidance if not. It supports configurable limits (max cycles, timeout, verbosity, branch, local-verify attempts) and handles many status codes with documented fix flows — applying fixes via MCP, applying locally and enhancing, rejecting and fixing from scratch, environment reruns, and throttling. It distinguishes environment/tooling failures from genuine code failures to avoid wasting fix budget, and includes git-safety guidance (stage specific files rather than 'git add -A' to avoid committing unrelated work or secrets). Some flows create commits and push, and can create empty commits to re-trigger CI.
It targets developers and teams using Nx and Nx Cloud who want automated CI monitoring and self-healing on their branches. Use cases include tracking CI status, auto-applying and verifying self-healing fixes, and recovering from throttled or failed pipeline attempts.
A workspace connected to Nx Cloud — it checks nx.json for nxCloudId or nxCloudAccessToken and exits with guidance if neither is present — plus the Nx Cloud MCP tools ci_information and update_self_healing_fix.
Some fix flows commit and push changes, and it can create empty commits to re-trigger CI. It follows git-safety guidance by staging specific files by name rather than using 'git add -A/.' to avoid committing unrelated work or secrets.
It spawns subagents to call one Nx Cloud MCP tool at a time and feeds the result into deterministic scripts (ci-poll-decide.mjs and ci-state-update.mjs) that return an action and manage budget gates, so decisions and state transitions are deterministic rather than ad hoc.
Defaults include --max-cycles 10, --timeout 120 minutes, --verbosity medium, auto-detected branch, --new-cipe-timeout 10, and --local-verify-attempts 3, all overridable via arguments.
It distinguishes environment/tooling failures (missing binaries, OOM, permission denied, network timeouts, disk exhaustion) from genuine code failures, and bails immediately on environment issues without consuming fix budget.
Quick Setup:
.claude/skills/Repository
nrwl/nx-ai-agents-config