The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to scaffold the Fumadocs site using npx, install project dependencies, and run the production build via pnpm build. These operations are performed within the output directory specified by the user.
[EXTERNAL_DOWNLOADS]: The skill downloads the create-fumadocs-app tool from the NPM registry and clones source material from GitHub repositories when a URL is provided as input. These are well-known services and the actions are central to the skill's functionality.
[INDIRECT_PROMPT_INJECTION]: The skill has an inherent vulnerability surface because it processes external source material (code, articles, and documentation) to generate a website. Malicious instructions embedded in the source material could theoretically attempt to influence the agent during the incremental authoring process.
Ingestion points: The scripts/inventory.py script and the incremental authoring loop described in SKILL.md (Step 6).
Boundary markers: The instructions do not explicitly define boundary markers or 'ignore' instructions when reading untrusted file contents.
Capability inventory: The skill utilizes shell command execution (via subprocess.run in scripts and pnpm build) and extensive file system write capabilities within the documentation output directory.
Sanitization: No explicit content sanitization or escaping logic is implemented in the provided Python scripts.

lov-repo2docs