The Agent Skills Directory

[SAFE] (SAFE): The skill consists of documentation and utility scripts intended for debugging. The techniques described (Root Cause Tracing, Defense-in-Depth, Condition-Based Waiting) are industry best practices for software engineering.
[Indirect Prompt Injection] (LOW): The skill's primary function involves ingesting and analyzing untrusted external data, such as error logs and stack traces, which creates a potential surface for indirect prompt injection.
Ingestion points: Error messages, stack traces, and logs processed in Phase 1 (SKILL.md) and root-cause-tracing.md.
Boundary markers: None present; the skill lacks explicit instructions to treat error messages as untrusted data or use delimiters.
Capability inventory: The skill includes file system searching, environment variable inspection (env), and test execution (npm test) via find-polluter.sh and SKILL.md.
Sanitization: No sanitization or filtering of external error data is implemented before analysis.
[COMMAND_EXECUTION] (SAFE): The script find-polluter.sh executes npm test. While this involves code execution, it is the expected behavior for a debugging utility designed to find failing tests. The script correctly quotes variables to prevent basic shell injection.

systematic-debugging