The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard development tools like git and gh to examine the repository state and retrieve pull request information. These operations are essential for identifying the scope of changes to be verified.\n- [DYNAMIC_EXECUTION]: The skill directs the agent to discover and invoke existing verifier skills and to create new skill definitions (.claude/skills/verify/SKILL.md) to persist successful build and launch recipes for future use.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted data including git diff outputs, PR descriptions, and the runtime output of the applications being tested. This ingestion of external data constitutes a potential attack surface.\n
Ingestion points: git diff, gh pr diff, and stdout/stderr from the verified application (SKILL.md).\n
Boundary markers: The instructions do not specify explicit boundary markers or sanitization steps for the ingested data.\n
Capability inventory: The skill has the ability to execute shell commands (git, ls, etc.), invoke other skills, and write files to the local workspace.\n
Sanitization: Not present.

verify