The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill's core functionality involves generating project-local shell scripts (e.g., start.sh, health.sh, stop.sh, compare.sh) and then instructing the agent to execute these scripts in Step 6 ('Run it once for real') to verify the setup.
[INDIRECT_PROMPT_INJECTION]: The generated skill is designed to ingest data from potentially untrusted external sources to define test expectations (answer keys).
Ingestion points: The skill reads PR bodies via gh pr view, issue descriptions, and project specification files found in the repository.
Boundary markers: The generated skill template does not include explicit delimiters or instructions to ignore embedded prompts within the ingested data.
Capability inventory: The agent has the ability to execute shell commands, run generated scripts, use curl, and perform git operations.
Sanitization: The skill does not implement sanitization or validation of the content retrieved from PRs or issues before using it to guide verification steps.
[EXTERNAL_DOWNLOADS]: The skill recommends the use of npx playwright install to set up browser-based verification environments. Playwright is a well-known and trusted automation service.
[COMMAND_EXECUTION]: The skill uses various standard developer tools and CLI utilities, including git, gh (GitHub CLI), curl, tmux, and docker, to map app surfaces and verify connectivity.

create-prove-it