parallel-search-setup

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides explicit instructions to mitigate risks associated with untrusted data retrieved from the web.
  • Ingestion points: The tools web_search and web_fetch ingest content from external websites.
  • Boundary markers: The skill instructs the agent to treat retrieved content as 'untrusted evidence'.
  • Capability inventory: The skill describes subprocess commands for MCP server registration and tool usage.
  • Sanitization: Instructions mandate that the agent must ignore instructions embedded in search results and avoid sending sensitive information to retrieved URLs.
  • [COMMAND_EXECUTION]: The skill provides CLI commands for users to register Parallel MCP servers. These commands use platform tools like claude mcp to connect to vendor-owned infrastructure at parallel.ai.
  • [CREDENTIALS_UNSAFE]: The skill implements best practices for secret management by forbidding the agent from requesting API keys in chat or searching local files for secrets, instead using the client's native secure storage or OAuth flows.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 03:43 PM