The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub PR status checks and external CI log URLs.
Ingestion points: Data is fetched from GitHub pull request status checks and associated targetUrl logs as described in the PR-Based Workflow section.
Boundary markers: None. The skill does not provide explicit delimiters or warnings to ignore embedded instructions in the external logs.
Capability inventory: The skill instructs the agent to use shell execution (Bash), file writing (Write), and HTTP requests (curl).
Sanitization: None. The instructions do not require the agent to sanitize or validate the $SUITE_NAME or test descriptions extracted from CI logs before interpolating them into shell commands such as pnpm dev $SUITE_NAME or playwright test -g "test name". This creates a potential command injection vector if the logs are maliciously crafted.
[COMMAND_EXECUTION]: The skill relies on several shell commands for environment management and test execution, including lsof, kill, pnpm, curl, and playwright. Specifically, it uses lsof -ti:3000 | xargs kill -9 to forcefully terminate processes occupying network ports.

triage-ci-flake