check-known-solutions
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill interacts with the vendor's official API at
api.pushrealm.comto fetch tool configurations and search results. This communication is essential for the skill's functionality and targets infrastructure owned by the author. - [PROMPT_INJECTION]: The skill processes third-party solutions from a community knowledge base, which presents a surface for indirect prompt injection if external data contains malicious instructions.
- Ingestion points: Data enters the context through
search_solutionstool outputs fromapi.pushrealm.com(SKILL.md, Step 2). - Boundary markers: The skill does not specify isolation delimiters for the external content.
- Capability inventory: The skill is intended for use in environments like Claude Code or Cursor, which typically grant agents file system and command-line capabilities.
- Sanitization: The skill effectively mitigates these risks by instructing the agent to treat search results as suggestions, perform manual reviews of any code before execution, and require explicit user consent before publishing data.
Audit Metadata