The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The update command directs users to fetch an installation script from the author's GitHub repository at https://raw.githubusercontent.com/odysseyalive/claude-enforcer/main/install. As this is a vendor-owned resource, it is classified as a functional update mechanism.
[REMOTE_CODE_EXECUTION]: Documentation for the update command provides a curl | bash command pattern for users to execute. This is intended for self-updating the tool from the official repository.
[COMMAND_EXECUTION]: The hooks and ledger command procedures involve creating local bash scripts and modifying the .claude/settings.local.json configuration file to wire these hooks into the environment. The skill also describes using commands like chmod +x to manage script permissions.
[PROMPT_INJECTION]: The skill processes other local files such as CLAUDE.md and various SKILL.md files to extract directives and suggest optimizations, which represents an indirect prompt injection surface.
Ingestion points: Reads and greps all .claude/skills/*/SKILL.md files and the project root CLAUDE.md.
Boundary markers: Uses structural headings like ## Directives and blockquote formatting to delimit user instructions.
Capability inventory: Utilizes TaskCreate for spawning sub-agents and Write/Edit for modifying skill files and system configurations.
Sanitization: The analysis does not identify explicit sanitization of input data to prevent the agent from being influenced by malicious instructions embedded in scanned project files.

skill-builder