The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill explicitly instructs the agent to read an external documentation file (https://www.ripple-ts.com/llms.txt) for "comprehensive reference".
Ingestion points: SKILL.md 'Documentation Links' and 'llms.txt' reference.
Boundary markers: Absent. The skill does not provide any delimiters or warnings to ignore potentially malicious instructions within the external file.
Capability inventory: The skill provides and encourages the use of terminal commands including npx degit, npm install, and npm run dev (subprocess execution).
Sanitization: Absent. There is no validation of the content fetched from the remote URL before it is processed by the agent.
[External Downloads] (MEDIUM): The skill promotes downloading templates from an unverified GitHub repository (Ripple-TS/ripple) and installing packages from NPM (ripple, @ripple-ts/vite-plugin).
The 'Ripple-TS' organization is not within the defined [TRUST-SCOPE-RULE], meaning these dependencies must be manually audited for safety.
[Command Execution] (MEDIUM): Provides executable shell patterns for project setup and running a development server.
Evidence: npx degit Ripple-TS/ripple/templates/basic my-app and npm run dev in SKILL.md.

ripple-ts