The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes go and golangci-lint to perform mechanical verification of linting rules. These are standard, well-known development tools used for their intended purpose.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the codebase in the form of code comments which are then processed by the agent during its analysis phase.
Ingestion points: Code comments following //nolint directives are extracted from Go source files via grep in SKILL.md.
Boundary markers: The skill defines a "Red/Blue/White" debate framework which provides a logical structure for the analysis, though it does not implement formal escaping for the ingested comment text.
Capability inventory: The skill uses subprocess execution to run the golangci-lint binary and performs file system reads.
Sanitization: No specific sanitization or filtering is applied to the justification comments before they are evaluated by the agent.

go-nolint-audit