The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands such as ls, find, cat, mkdir, ln, mv, and rmdir, and utilizes the GitHub CLI (gh) to inspect and update local project configurations. These operations are primarily scoped to the .claude/ directory and require explicit user consent for any destructive changes.
[EXTERNAL_DOWNLOADS]: The skill fetches agent skills from the author's repository (Fandhe-AI/agent-cli-skills) using the skills CLI. The instructions strictly enforce the use of a pinned version (1.5.23) of the CLI tool to mitigate supply chain risks, and included regression tests verify this pinning.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes content from local configuration files (Agents, Rules, CLAUDE.md) and GitHub API metadata. Malicious content within these external sources could theoretically influence the agent's diagnosis or output. The skill mitigates this through a mandatory user approval step (Step 3) and specific instructions for the agent to review all differences with the user.

update-claude