The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates several shell scripts and installs them into the local repository's configuration directory.
Generates guard.sh, format.sh, stop-quality-gate.sh, and other utility scripts in .claude/hooks/.
Configures these scripts as automated hooks in .claude/settings.json to trigger during agent events like tool execution, file writes, and session termination.
[EXTERNAL_DOWNLOADS]: Recommends the installation of several well-known third-party tools to support analysis and project security.
Mentions jq, gitleaks, semgrep, and the GitHub CLI (gh).
References an external skill from the same vendor (squirrelsoft-dev/agent-skills@workflow) for ongoing workflow automation.
[PROMPT_INJECTION]: The skill ingests untrusted data from the local repository to facilitate automated configuration.
Scripts like analyze-conventions.sh and analyze-git.sh sample source files and git history.
The analysis results are used to generate natural language rules in .claude/rules/, creating an indirect injection surface where content in the repository could influence the analysis logic, although findings are presented for human review before being applied.

brownfield