The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches documentation and project metadata from the project's official GitHub repository via scripts/fetch-docs.sh. This is used to maintain up-to-date guidance for the agent and targets a well-known service.
[COMMAND_EXECUTION]: Provides instructions for standard development operations, including git clone, npm install, and npm run dev. These are routine for the skill's developer-focused use case.
[PROMPT_INJECTION]: The skill manages an indirect prompt injection surface where external data is ingested into the agent context.
Ingestion points: Remote README and configuration files are retrieved via scripts/fetch-docs.sh as documented in SKILL.md.
Boundary markers: Present. The skill includes explicit instructions in the 'Best practices' section to treat external content as untrusted and ignore embedded instructions.
Capability inventory: The skill has access to Bash, Write, Edit, Read, Grep, and Glob across its referenced scripts and instructions.
Sanitization: Instructions mandate summarizing external content and warn against executing instructions contained within it, relying on agent reasoning to mitigate malicious content.

lapian-notes