composio-mcp

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill defines a COMPOSIO_REMOTE_BASH_TOOL which permits the execution of shell commands such as jq, awk, sed, and grep in a remote sandbox environment provided by the vendor.
  • [REMOTE_CODE_EXECUTION]: The COMPOSIO_REMOTE_WORKBENCH tool allows for the execution of Python scripts and the orchestration of multi-tool chains within a persistent remote Jupyter notebook.
  • [PROMPT_INJECTION]: The skill facilitates the processing of data from a large number of external services, which introduces a surface for indirect prompt injection.
  • Ingestion points: Untrusted data enters the agent context through tool outputs when fetching content from external apps like Gmail, Slack, or GitHub (referenced in SKILL.md).
  • Boundary markers: There are no explicit instructions or delimiters defined to isolate or sanitize instructions that may be embedded within the retrieved data.
  • Capability inventory: The skill provides high-privilege capabilities including remote shell execution, Python scripting, and the ability to trigger actions across 1000+ integrated services (referenced in SKILL.md).
  • Sanitization: The documentation does not specify any sanitization or validation steps for the content ingested from external connectors.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 04:42 PM
Security Audit — agent-trust-hub — composio-mcp