The Agent Skills Directory

[COMMAND_EXECUTION]: The skill inherently performs dynamic code execution as part of its core functionality.
Evidence: Step 6.1 through 6.3 involves writing regression tests and running the project's full test suite to verify fixes. Additionally, Step 4 instructs a sub-agent to reproduce the bug, which typically requires executing project code.
Context: This behavior is expected for a TDD-based debugging tool, but users should be aware that running tests on untrusted codebases could trigger malicious side effects if the project itself is compromised.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through external data ingestion.
Ingestion points: Step 1 accepts bug descriptions from prompt text, external files (e.g., @issues/bug-123.md), and the current conversation context (review reports).
Boundary markers: The skill does not implement explicit delimiters or "ignore instructions" warnings when passing these inputs to the sub-agent in Step 4.
Capability inventory: The agent has permissions to read/write files, execute shell commands (test suites), and perform git commits.
Sanitization: No sanitization or validation of the input bug reports is described before processing.
Risk: A maliciously crafted bug report could attempt to influence the sub-agent's diagnosis or the final code fix instructions.

fixbug