CI Triage x402 — Developer Reliability Tools

Parse a CI, build, compiler, dependency, container, lint, or test log into a secret-redacted failure type, primary error, useful trace, fix category, and confidence score. Use it to analyze build logs and identify the first actionable root-cause signal.

Generate a stable, secret-free fingerprint from a build or test failure so agents can group duplicate CI errors across changing line numbers, timestamps, paths, identifiers, and retry attempts.

Strip ANSI and runner noise while redacting GitHub tokens, AWS access keys, bearer credentials, JWTs, API keys, passwords, and common secret assignments from CI/CD logs before an agent stores, shares, or analyzes them.

Analyze a public GitHub Actions run URL using failed-job metadata and check annotations. Return the failing job and step, root-cause classification, redacted evidence, and repair guidance without requesting a GitHub token or pretending public raw logs are accessible.

Analyze as many as twenty-five CI, compiler, dependency, container, lint, and test logs in one paid call. Return a consistent secret-redacted diagnosis for every input so an agent can process a failed build matrix efficiently.

Turn one failing build, test, compiler, dependency, lint, or container log into an ordered diagnosis with likely root cause, redacted evidence, a focused rerun command, concrete repair steps, and a stable failure fingerprint.

Run redaction, root-cause triage, duplicate fingerprinting, focused rerun selection, repair planning, and reviewer-ready Markdown handoff in one call. This is the complete build-failure workflow for an autonomous coding agent.

Parse bounded JUnit XML from pytest, Jest, Maven, Gradle, or other test runners. Return suite totals, pass rate, duration, failed test names, failure messages, and the files or classes that need attention without executing embedded content.

Normalize a SARIF 2.x report from CodeQL, Semgrep, Trivy, or another scanner into severity counts, top rules, prioritized findings, file locations, and concise messages suitable for an agent's remediation queue.

Identify npm and Node.js failure modes such as ERESOLVE peer conflicts, ETARGET versions, engine mismatches, node-gyp builds, permission errors, missing files, lockfile drift, and registry/network failures, then return safe recovery commands.

Parse pip and Python packaging failures including ResolutionImpossible, incompatible requirements, missing distributions, wheel build failures, Python-version constraints, SSL/index errors, and broken environments, then return focused verification and repair steps.

Parse a GitHub Actions workflow as inert YAML and report structural errors, missing jobs or runners, malformed steps, dangerous pull_request_target patterns, mutable action references, overly broad permissions, and common expression mistakes.

Classify Docker and BuildKit errors such as missing COPY context, failed RUN commands, unavailable image manifests, apt failures, permissions, network resolution, platform mismatches, disk exhaustion, and cache problems, with focused rebuild commands.

Extract the first actionable TypeScript diagnostic, error code, file, line, and message; group repeated diagnostics and return a focused tsc command plus repair guidance for type, module, configuration, and declaration failures.

Extract the failed pytest node id, assertion or exception, relevant Python frames, likely cause category, focused single-test rerun command, and practical repair steps from a pytest failure log.

Extract the failing Jest suite and test, expected-versus-received assertion, JavaScript or TypeScript frames, focused rerun command, and repair guidance from Jest, Vitest, or compatible test output.

Normalize stylish, compact, and common ESLint output into error and warning counts, affected files, first actionable rule violations, focused rerun command, and whether safe automatic fixing may be available.

Parse go test, go vet, and compiler output into failed package or test, panic or compiler message, relevant Go frames, focused rerun command, and repair guidance for tests, modules, races, and compilation errors.

Extract Rust compiler error codes, source locations, Cargo dependency failures, borrow-checker diagnostics, failed tests, and panics; return focused cargo commands and evidence-linked repair guidance.

Create a retry-stable fingerprint while detecting timing, race, randomness, ordering, shared-state, port, resource, and intermittent-network signals that often distinguish flaky tests from deterministic failures.

Infer a safe, non-executed rerun command from pytest, Jest, Vitest, ESLint, TypeScript, npm, pip, Docker, Go, Rust, Maven, or Gradle failure output so an agent can reproduce the smallest failing unit.

Parse LCOV, Cobertura XML, coverage.py JSON, Istanbul JSON summaries, and common terminal coverage tables. Return line, branch, function, and statement percentages plus threshold status in one normalized response.

Validate dotenv syntax without returning secret values. Report invalid or duplicate keys, malformed assignments, risky committed-secret key names, quoting problems, export usage, and line numbers while keeping all values private.

Parse Python, JavaScript, TypeScript, Java, Go, and Rust stack traces into language, exception, message, and normalized frame objects with function, file, line, and column fields for downstream debugging agents.

Convert a raw failing build or test log into concise secret-redacted Markdown containing the root cause, evidence, fingerprint, reproduction command, repair checklist, and confidence for an issue, pull request, or human reviewer.

Analyze up to twenty-five build or test logs, group retries and duplicate failures by stable secret-free fingerprints, and return cluster counts, representative messages, failure types, and the highest-frequency root cause.

Compare a previous and current build or test failure after secret redaction and noise normalization. Report whether the root-cause fingerprint changed, new and removed error evidence, similarity, and diagnosis before and after.

Normalize GitHub Checks or compatible annotation objects into severity counts, affected files, top rules, prioritized evidence, and a concise failure summary. Secret-like content is redacted before it is returned.

Auto-detect JUnit XML, SARIF JSON, LCOV, coverage tables, pytest, Jest, ESLint, Go, Rust, and generic CI output, then return a consistent report type, status, totals, failures, diagnosis, and source-format metadata.

Detect likely GitHub tokens, AWS keys, bearer credentials, JWTs, private-key blocks, database URLs, API keys, passwords, and secret assignments in a CI log. Return only types, counts, line numbers, and redacted snippets—never the detected values.