The Agent Skills Directory

[SAFE]: The skill consists of instructional content for performing web research using built-in agent tools. No malicious patterns or security risks were identified during the analysis.
[PROMPT_INJECTION]: The skill was evaluated for potential indirect prompt injection vulnerabilities as it involves processing large amounts of external web content. 1. Ingestion points: Untrusted data enters the agent context via the webfetch tool calls described in the Research Workflow and Research Templates (SKILL.md). 2. Boundary markers: The instructions do not define specific delimiters or warnings to ignore instructions embedded in the fetched web content. 3. Capability inventory: The skill's capabilities are limited to searching, fetching, and synthesizing text. It does not perform file writes, network exfiltration, or command execution. 4. Sanitization: There are no explicit content filtering or sanitization steps mentioned in the workflow. While these factors identify a potential attack surface, it is consistent with the primary research purpose and the risks are mitigated by the agent's restricted toolset.

deep-research