connect

Pass

Audited by Gen Agent Trust Hub on Jun 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill guides users to install several packages including 'composio', '@composio/core', 'claude-agent-sdk', 'openai-agents', and 'composio-langchain'. These are official libraries for a well-known tool-routing service.
  • [PROMPT_INJECTION]: The skill facilitates indirect prompt injection by design, as it allows an agent to process data from external sources and perform actions based on it.
  • Ingestion points: Data enters the agent context from connected apps such as Gmail, Slack, and GitHub via the Tool Router.
  • Boundary markers: The skill does not provide explicit delimiters or warnings to the agent to ignore instructions embedded within the data retrieved from external apps.
  • Capability inventory: The skill enables high-impact capabilities including sending emails, posting chat messages, updating databases, and creating repository issues.
  • Sanitization: There is no evidence of input validation or sanitization for content received from external services before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 20, 2026, 03:09 PM
Security Audit — agent-trust-hub — connect