CI Failure Triage Lite

    1

    Fast triage of failed CI logs to identify root causes, decode exit codes, and classify infrastructure vs. code errors.

    Free

    1 useSecurity scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    CI Failure Triage Lite

    Example session with this skill installed

    My GitHub Actions job just failed with 'Process completed with exit code 137'. Here is the tail of the log: [Log shows a heavy test suite followed by a sudden termination message]

    • Read your context and instructions
    • Compiled the ci failure triage

    I have analyzed the provided log snippet. The failure is not a test bug, but an external termination.

    CI Triage Report

    MetricDetails
    Verdictinfrastructure-or-environment
    ConfidenceHigh
    Failing Stepnpm test
    Exit Code137 (SIGKILL)
    Tests RanPartially

    Evidence

    • Line 842: Killed
    • Line 843: Process completed with exit code 137.

    Analysis

    The process was terminated by the host operating system, likely the GitHub Actions runner. Exit code 137 specifically indicates a SIGKILL (128+9). This occurs most frequently when the runner runs out of memory (OOM) and the OOM killer terminates the most resource-intensive process. The log shows 42 tests passing immediately before the kill, suggesting cumulative memory pressure rather than a specific logic error.

    Recommended Experiments

    Check Runner Resources: Inspect the GitHub Actions runner stats to see if memory peaked at 7GB (for standard runners).
    2. Reduce Parallelism: Try running npm test -- --maxWorkers=2 to see if lower concurrency prevents the OOM kill.
    3.

    Memory Profile: If the failure persists, run the suite locally with a memory profiler to identify leaks in the test setup.

    Next steps

    • Verify if this suite recently increased in size or if the runner type was changed.
    • Check the dmesg or system logs if using a self-hosted runner to confirm OOM killer activity.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Stop wasting hours scrolling through thousands of lines of CI noise. When a pipeline turns red, this skill identifies the root cause by scanning for the first true error and decoding cryptic exit codes. It separates infrastructure meltdowns from genuine code regressions and flaky test suites.

    What it does

    • Exit code decoding translates codes like 137, 143, and 139 into human-readable causes like OOM kills or signal timeouts.
    • Root cause isolation identifies the first point of failure in a log to distinguish the primary error from cascading fallout.
    • Failure classification categorizes issues into infrastructure, network, dependency drift, or deterministic breakage.
    • Evidence extraction quotes exact log lines and line numbers to support every diagnostic claim.
    • Experiment recommendation suggests the single most cost-effective action to confirm a diagnosis.

    How it works

    1. Intake the raw CI log or a link to the failing job output from GitHub Actions, GitLab CI, or Jenkins.
    2. Analysis of the failing step, exit codes, and the sequence of logs to find where the process first deviated.
    3. Verdict generation providing a confidence score (High/Medium/Low) and a specific failure label.

    Frameworks & tools

    Works with any CI provider including GitHub Actions, GitLab CI, Jenkins, and CircleCI. Compatible with logs from any language or test framework, such as Jest, Pytest, Go test, or JUnit.

    Why this beats prompting it yourself

    Generic prompts often hallucinate based on the last few lines of a log, which are usually just cleanup errors. This skill follows a strict "first error" protocol and uses a pre-defined reference library for Unix exit signals, ensuring infrastructure issues aren't misdiagnosed as code bugs.

    Use cases

    • Triage intermittent failures to determine if a "flaky" test is actually a resource exhaustion issue.
    • Debug environment drift when a build passes locally but fails in CI due to dependency resolution errors.
    • Analyze OOM kills by identifying SIGKILL signatures in silent or truncated process logs.

    Known limitations

    Requires log text to function. Cannot access private repositories or execute code to verify fixes. One log cannot statistically prove flakiness on its own.

    Requires raw log text. Cannot access private repos or run code. Cannot prove flakiness from a single run.

    How to use

    1. 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. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    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

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Git

    Permissions

    No special permissions declared or detected

    Listed1 month ago
    Updated1 month ago

    What's inside

    Frequently Asked Questions

    Unlike basic prompts that focus on the end of a log, this skill uses a specialized protocol to scan for the 'first true error' and decodes specific Unix exit signals like 137 (OOM) or 139 (Segfault) that generic models often miss.

    This skill is framework-agnostic and works with any text-based logs from GitHub Actions, GitLab CI, Jenkins, and CircleCI, regardless of whether your code is in Python, JavaScript, Go, or Java.

    The skill requires the raw text output of your failing CI job; it does not require direct access to your private repositories or sensitive source code to perform its analysis.

    You receive the specialized CI Failure Triage Lite skill logic, which can be integrated into your preferred AI agent environment to begin analyzing logs immediately.

    While it can identify the symptoms of a flaky test (like a signal timeout), a single log analysis cannot statistically prove flakiness; it is best used to rule out infrastructure and environment issues first.

    New in Testing & Debugging