The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local system commands to perform its core functions.\n
Helper scripts like scripts/lib/index.mjs and scripts/context-verify.mjs use child_process.execSync to run git commands (such as git rev-parse and git diff) for auditing code history and verifying documentation provenance.\n
The scripts/change-archive.mjs script uses the zip utility to package completed change workspaces.\n
Security analysis confirms that inputs to these commands, such as Git SHAs and change IDs, are validated using strict regular expressions or sanitized during generation to prevent shell injection vulnerabilities.\n- [SAFE]: The skill operates entirely within the local environment and implements several security best practices.\n
Human-in-the-Loop: The workflow explicitly requires user confirmation at multiple stages (selection gate, implement gate, docs gate) before proceeding with code changes or archiving, preventing autonomous malicious actions.\n
Local Execution: All script operations are performed on the local filesystem without external network requests or downloads from untrusted sources.\n
Least Privilege: The scripts utilize standard Node.js APIs and focused shell commands for specific tasks like Git metadata retrieval, rather than requesting broad administrative shell access.

refactor