The Agent Skills Directory

[SAFE]: The skill implements a standard software development workflow for test maintenance. No evidence of malicious behavior, unauthorized network access, or hardcoded credentials was found.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it involves reading and interpreting output from external test commands. * Ingestion points: Test failure logs and error messages generated by running tests as described in SKILL.md. * Boundary markers: Not present; the agent is instructed to read the output directly. * Capability inventory: The skill utilizes shell command execution (make, pytest, git) and file editing tools to implement fixes. * Sanitization: No explicit sanitization or filtering of test output is mentioned before the agent processes it.

test-fixing