The Agent Skills Directory

[SAFE]: The skill enforces a high-integrity analysis process by requiring that all information be categorized as VERIFIED, INFERRED, or UNKNOWN. It provides explicit security instructions to never expose secret values and to treat external documentation as hypotheses until confirmed by the actual code, which mitigates risks associated with misleading or malicious codebase documentation.
[SAFE]: The included Python script (scripts/validate_scenarios.py) is a developer utility for static contract validation. It only reads the text content of local markdown files within the skill repository to ensure required analysis templates are present. It does not perform network requests, execute external commands, or access sensitive system paths.
[SAFE]: The skill manages the potential for indirect prompt injection by instructing the agent to cross-reference multiple evidence sources (code, configuration, migrations, and tests) and to explicitly state the confidence levels of its findings, thereby reducing the likelihood of the agent being misled by adversarial content within an analyzed repository.

codebase-understanding