The Agent Skills Directory

[SAFE]: The skill is a standard documentation retrieval tool that operates within the expected scope for a product assistant. No malicious behaviors, obfuscation, or credential harvesting were identified.
[EXTERNAL_DOWNLOADS]: The skill fetches documentation from prismatic.io and code examples from github.com/prismatic-io. These are the official domains and repositories of the skill's author (the vendor) and are well-known, trusted services.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from documentation pages and GitHub repositories, which is its primary function.
Ingestion points: Documentation markdown content from prismatic.io/docs/ and raw code files from github.com/prismatic-io/examples (as described in SKILL.md and references/example-code.md).
Boundary markers: None explicitly defined in the instructions for the retrieved content.
Capability inventory: The skill uses standard web search and fetch tools to retrieve information; no dangerous capabilities such as local file system writes, subprocess execution, or unauthorized network operations are present in the provided scripts or instructions.
Sanitization: The skill explicitly instructs the agent to fetch .md versions of documentation pages to strip out HTML, CSS, and JavaScript, which reduces the surface area for injection attacks.

prismatic-docs