The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git rev-parse and git ls-files to identify the repository root and the list of tracked files for scanning. This is standard behavior for a repository auditing tool.
[DATA_EXPOSURE]: The skill's primary purpose is identifying existing data exposure (leaked secrets). It generates a local report at .claude/reports/ containing these findings for user review. It does not attempt to send this data to external servers.
[SAFE]: All identified tools (Read, Agent, Write, Bash, AskUserQuestion) are used in alignment with the skill's stated purpose of auditing and remediating local code issues.
[SAFE]: The sub-agent prompts are well-defined, instructing the agents to perform read-only analysis and return structured JSON, preventing uncontrolled behavior during the parallel scan phase.

secrets-scan