The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data (project requirements briefs or site content) and has the capability to write to the local file system and perform web operations. This creates a surface for indirect prompt injection attacks where malicious instructions embedded in a project brief could influence the agent's behavior.
Ingestion points: The skill instructions explicitly require reading an external "project requirements brief or description document" in the Action Plan and Prerequisites.
Boundary markers: The instructions lack any boundary markers, delimiters, or explicit instructions for the agent to ignore or isolate commands found within the ingested documentation.
Capability inventory: The skill is granted Read, Write, and web_fetch tools, allowing it to interact with the file system and external networks.
Sanitization: There are no instructions provided to sanitize, validate, or filter the content extracted from external briefs before it is used to generate the final documentation or perform further actions.

site-md