- Home
- Skills
- DevOps & Cloud
- CI Failure Explainer
Works with the AI Agents you already use
CI Failure Explainer
A CI doctor for your coding agent: turns a huge red GitHub Actions log into a verified root cause and the smallest fix, with secrets redacted locally.
Free
CI Failure Explainer
Example session with this skill installed
Why did CI fail on my PR?
- Read your context and instructions
- Compiled the ci failure explainer
I ran the triage on the failed run (latest failed GitHub Actions run on this branch). The raw log was redacted and shrunk to a short digest before I read it.
CI failure: CI › test (3.12) › Run pip install -r requirements.txt
Verdict: ledger/client.py now imports httpx, but httpx isn't in requirements.txt, so pytest can't even collect tests/test_client.py.
| Field | Value |
|---|---|
| Category | dependency |
| Confidence | high |
| Likely flaky | no |
| Known pattern | PY-03 · ModuleNotFoundError in CI |
Evidence
- Log:
E ModuleNotFoundError: No module named 'httpx' - Change:
ledger/client.py:7addsimport httpxin this PR;requirements.txtis unchanged
Fix
Add httpx>=0.27 to requirements.txt.
Verify
python -m venv /tmp/v && /tmp/v/bin/pip install -r requirements.txt && /tmp/v/bin/pytest -x
If this isn't it
If httpx is already listed, CI may be installing a different extras group. Compare the workflow's pip command with your local one.
Want me to apply this fix?
Connects securely to your tools. The creator never sees your data.
About this skill
CI failures waste hours of developer time on log archaeology and secret redaction. This skill automates the triage of broken GitHub Actions, GitLab, Jenkins and CircleCI runs by turning massive, noisy logs into a concise markdown brief with a verified root cause and a minimal fix.
What it does
- Automated triage fetches failed GitHub Actions runs or parses pasted logs to build a compact diagnostic brief.
- Log compression shrinks logs by 10-100x into a 150-line digest focusing only on failure signals and errors.
- Secret redaction automatically scrubs credentials from logs before processing to prevent accidental leakage in chat history.
- Pattern matching compares failures against a library of 160 known error patterns for Flutter, Node, Python, Docker, Go, Rust, JVM, Ruby, .NET, PHP and GitHub Actions.
- Root cause analysis correlates failures with recent git diffs to isolate whether the breakage stems from code changes, toolchain drift, or infrastructure.
How it works
- Run
ci_doctor.pywith a PR number, run ID, or local log file to generate a triage brief and bundle. - Review the digest and detected stack patterns to identify the failing step and reproduction command.
- Use the diagnostic method to verify the failure locally using the generated
repro.shscript. - Apply the recommended minimal fix and verify the build passes before committing.
Frameworks & tools
The skill requires Python 3.9+ and the GitHub CLI (gh) for remote fetching. It includes specialized reference libraries for Node.js, TypeScript, Python, Go, Rust, JVM, Flutter, Docker, Ruby/Rails, .NET/C#, PHP, and GitHub Actions, GitLab and Jenkins logs.
Why this beats prompting it yourself
Generic LLMs often struggle with massive CI logs, hitting context limits or missing subtle version mismatches. This tool uses pre-processing scripts to redact secrets and distill logs, providing the agent with high-signal data and a structured diagnostic framework that standard prompts lack.
Use cases
- Fixing PR breakages by identifying why a check is red without manual log scrolling.
- Debugging toolchain drift when a previously passing build fails due to environment changes.
- Triaging flaky tests using historical pass/fail signals and specialized flakiness references.
- Safe log analysis for security-sensitive projects where raw logs cannot be shared directly with an AI.
Known limitations
Fetching remote runs requires an authenticated GitHub CLI session. The skill is optimized for Linux and macOS, requiring WSL or Git Bash for Windows users.
Open source
Free and MIT-licensed. Source code, issues and updates: https://github.com/hasanat-ahmad/CI-Failure-Explainer
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for DevOps & Cloud.
Requires Python 3.9+ and GitHub CLI for remote fetches. Windows requires WSL or Git Bash. Log processing ignores lines flagged as potential injection attempts.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 13 days ago
- Free with an Agensi account
- Accounts
GitHub
- Software
Python 3.9+
GitHub CLI
Gittsc
Node.jsprismaeslintprettierjestvitestPlaywright
Permissions
Allowed Hosts
Runs its own Python scripts in the terminal. Network access is only used by the GitHub CLI (gh) to download failed run logs; pasted logs need no network. It writes a temporary digest and a reviewable repro.sh. Optional env vars CI_EXPLAINER_GH_RETRIES and CI_EXPLAINER_RETRY_DELAY tune retries. Logs are redacted locally before the agent reads them.