humanloop
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
@useclawlink/clitool to interact with Humanloop. It executes commands for logging in, connecting services, and performing actions such as creating or deleting projects. These are standard operations for a CLI-based integration tool. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from Humanloop projects and experiments which could theoretically contain instructions embedded in data.
- Ingestion points: Data is ingested through actions that list projects, sessions, or experiments (e.g.,
humanloop_list_sessions,humanloop_list_experiments). - Boundary markers: The instructions explicitly state to "confirm with the user before any write," acting as a human-in-the-loop safeguard.
- Capability inventory: The skill can execute shell commands via the ClawLink CLI to read or modify Humanloop data.
- Sanitization: Relies on the agent's ability to interpret CLI output and the requirement for user confirmation for writes.
Audit Metadata