The Agent Skills Directory

[SAFE]: The skill operates entirely through local Node.js scripts that manage project documentation and change tracking. The implementation uses standard Node.js built-in modules (fs, path, child_process, util) for file system and process management, with no external dependencies.
[COMMAND_EXECUTION]: The helper scripts utilize child_process.execSync to perform Git operations (git rev-parse, git diff) and system tasks like archiving files with zip. These commands are strictly for local project maintenance and follow the skill's defined purpose.
[PROMPT_INJECTION]: The skill defines a 'challenge protocol' and an adversarial stance to ensure technical rigor during the documentation process. These instructions are legitimate behavioral guidelines and do not contain patterns designed to bypass safety filters or exfiltrate data.
[INDIRECT_PROMPT_INJECTION]: The workflow involves an automated triage where a subagent reads CONTEXT.md files to identify inaccuracies. While processing user-provided documentation introduces an indirect injection surface, the risk is managed by the specific analytical nature of the task and the lack of high-privilege execution paths triggered by the document content.

verify