The Agent Skills Directory

[NO_CODE]: The skill consists entirely of natural language guidelines and structural templates for research workflows. No scripts, executables, or code-based configurations are included.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources during the research process, which is a standard surface for indirect prompt injection. 1. Ingestion points: Steps 2 and 3 involve identifying and exploring external sources, expert opinions, and technical details from the web. 2. Boundary markers: No explicit delimiters or specialized safety instructions are provided to separate external content from the agent's instructions. 3. Capability inventory: The workflow relies on the agent's ability to search the internet and read external content. 4. Sanitization: Step 4 (Cross-reference and Validate) provides qualitative checks like verifying facts and assessing source credibility to mitigate risks.
[SAFE]: The metadata, description, and instructions are consistent with the stated purpose. The content does not contain any hidden instructions, privilege escalation, or persistence mechanisms.

deep-researcher