integrate
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references standard SDK and CLI packages for the Olakai platform. Evidence: installation of '@olakai/sdk', 'olakai-sdk', and 'olakai-cli' via npm and pip.
- [DATA_EXFILTRATION]: The skill is designed to transmit interaction data (prompts, responses, and metadata) to the Olakai service for monitoring. This is the primary intended function of the integration.
- [COMMAND_EXECUTION]: The skill provides documentation for CLI commands used to configure agents and validate data flow, such as 'olakai login' and 'olakai activity list'.
- [PROMPT_INJECTION]: The skill establishes a data ingestion surface for untrusted LLM interactions. 1. Ingestion points: 'userMessage' and 'aiResponse' variables in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Data is transmitted to the Olakai API. 4. Sanitization: None; data is passed raw to the monitoring service as required for accurate observability.
Audit Metadata