monitor-ci

Monitors Nx Cloud CI pipelines and orchestrates self-healing fixes via a polling subagent.

3|Updated May 14, 2026
One-click install
npx skills add https://github.com/jbadeau/fabster --skill monitor-ci-jbadeau
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: monitor-ci
Source: https://github.com/jbadeau/fabster/tree/main/.cursor/skills/monitor-ci
Command: npx skills add https://github.com/jbadeau/fabster --skill monitor-ci-jbadeau

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Watching CI pipelines manually wastes developer time, and fixing broken builds requires constant context switching. This Skill automates monitoring of Nx Cloud CI executions and coordinates self-healing fixes so failures are resolved without manual polling. ## Core Features & Use Cases - Subagent-Based Polling: Spawns a dedicated ci-monitor-subagent to poll CI status in the background, keeping the main agent context clean. - Self-Healing Fix Orchestration: Applies, rejects, or locally enhances AI-generated fixes from Nx Cloud self-healing based on verification status. - Local Verification Flow: Runs failed tasks locally in parallel before applying fixes, with configurable retry budgets and commit/push automation. - Use Case: After pushing a feature branch, ask the agent to watch CI; it detects a failing lint task, applies the verified self-healing fix via MCP, waits for the new CI attempt, and reports success. ## Quick Start Ask the agent to monitor CI for the current branch and automatically handle any self-healing fixes until the pipeline passes.

Frequently Asked Questions about monitor-ci

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I monitor CI status for my branch automatically?

Invoke the monitor-ci skill with your branch, and it spawns a background subagent that polls Nx Cloud CI status. The main agent relays status updates and handles fixes based on the returned status, without manual polling.

How does Nx Cloud self-healing fix application work?

When a fix is available, the skill compares failed and verified task IDs. Fully verified fixes are applied via the MCP update_self_healing_fix tool, while unverified non-e2e tasks go through local verification before applying.

Can I use gh or glab CLI to watch CI instead?

No, the skill explicitly prohibits CI provider CLI watch flags because they bypass Nx Cloud self-healing. Provider CLIs are only acceptable for one-time read-only status checks, never continuous monitoring.

What happens when no new CI attempt appears after a push?

The subagent returns a no_new_cipe status after the configured timeout. The skill reports the pre-CI failure and, if auto-fix-workflow is enabled, attempts a lockfile update and push to retrigger CI.

Why does the skill require Nx Cloud to be connected?

Self-healing fixes and CI attempt tracking depend on Nx Cloud APIs. The skill checks nx.json for nxCloudId or nxCloudAccessToken before starting and exits with a connection message if neither exists.

What are the limitations of the monitoring loop?

Monitoring stops at the max agent-initiated cycle count (default 10), the overall timeout (default 120 minutes), or after 3 consecutive no-progress iterations. Human-pushed cycles do not count against the cycle limit.