The Agent Skills Directory

[SAFE]: The skill is well-defined and operates within a limited security context. It specifically allows only read-oriented tools (Read, Grep, Glob) and specific version control queries (git log, git diff).
[SAFE]: Explicit safety instructions are provided to the agent, directing it to avoid production environments, exclude sensitive information from examples, and refrain from modifying the codebase.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided content ($ARGUMENTS) such as bug descriptions and issue URLs. While this represents an attack surface for indirect prompt injection, the risk is mitigated by the lack of network access tools and the limited, read-only nature of the available commands. No specific malicious patterns were found in the handling of this data.

bug-reproduce