Diagnose a failing or recently-failed e2e smoke test in this repo by walking through the forensics bundle (summary.json + prompt.txt + response.txt + snapshot.txt + screenshot.png) following the procedure in e2e/runbook.md. Accepts a GitHub Actions run URL, a local artifact zip, or an unpacked bundle directory. Identifies whether the failure is from app version drift, selector drift, an auth challenge, a Cloudflare interstitial, an assertion mismatch, or something else, and points at the likely fix. Use this skill whenever the user mentions an e2e test failure, smoke test failure, GHA failure for the claude-desktop or chatgpt-web jobs, a red CI run, a failing forensics bundle, or asks to investigate a failed run. Also use when invoked as `/triage-e2e-failure <run-url-or-path>`.
96%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
The required workflow downloads a specific user-selected GHA run artifact or local bundle and reads its `prompt.txt`, `response.txt`, and `snapshot.txt`, which are outputs of live-client e2e runs on the team's own accounts and repo-owned fixtures, so outsider text reaches the agent only if the agent first selects a specific run and only indirectly through the tested client's output.
boarlabsxyz/awesome-mcp
85cee7a
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.