skills/ahgraber/skills/mcp-research/Gen Agent Trust Hub

mcp-research

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMPROMPT_INJECTIONNO_CODE
Full Analysis
  • [Indirect Prompt Injection] (MEDIUM): The skill's primary function is to ingest untrusted data from the open web and external documentation tools, creating an attack surface for instructions embedded in processed data.
  • Ingestion points: Technical data is retrieved via tools such as mcp__exa__web_search_exa, mcp__jina__search_web, and mcp__jina__read_url as described in SKILL.md.
  • Boundary markers: There are no explicit instructions or delimiters defined to separate retrieved content from system instructions or to warn the agent about embedded commands.
  • Capability inventory: The skill is intended to support 'third-party integration work' and 'framework-specific debugging,' which typically involves the agent writing or modifying code based on the retrieved information.
  • Sanitization: No sanitization or validation steps are defined for the retrieved data before it is synthesized for the user.
  • [NO_CODE] (SAFE): The skill consists entirely of markdown instructions and metadata with no executable scripts or dependency files included in the provided package.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 02:48 AM