The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Playwright trace files, including console logs and error messages.
Ingestion points: trace.trace and trace.network files, and GitHub Actions artifacts.
Capability inventory: The skill uses child_process to execute extraction and summary scripts, and interacts with the filesystem and the gh CLI.
Sanitization: Instructions in rules/console-and-errors.md and rules/action-timing.md direct the agent to quote log messages and error narratives verbatim, creating a surface for indirect prompt injection if the trace contains malicious instructions.
Boundary markers: There are no explicit instructions to escape or sanitize this verbatim content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on local Node.js scripts (fetch-gh-run.mjs, trace-extract.mjs) that execute system-level commands.
Execution method: Uses child_process.execFileSync and spawnSync to invoke gh, unzip, and python3.
Purpose: Used for downloading artifacts from GitHub Actions and extracting ZIP archives.
Input flow: While arguments are passed as arrays (mitigating shell injection), the skill relies on the presence and integrity of these external tools on the host system.
[EXTERNAL_DOWNLOADS]: Fetches test artifacts and metadata from GitHub Actions.
Target: github.com.
Method: Uses the gh CLI (gh api and gh run download) to retrieve artifacts based on user-provided run URLs.

playwright-trace-analyzer