kelly-agent-observability
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a legitimate observability dashboard that operates on simulated data. No evidence of malicious behavior, obfuscation, or unauthorized access was found.
- [COMMAND_EXECUTION]: The skill provides a script (generate_fleet_data.mjs) for seeding the database with mock data. These commands are for initialization purposes and do not execute unauthorized code.
- [DATA_EXFILTRATION]: Data management is handled through a Busabase client. Credentials are managed via OAuth or environment variables and stored in standard local paths (~/.busabase/airapps), which is a legitimate part of the application workflow.
- [PROMPT_INJECTION]: The skill ingests user notes into a 'handoffs' database.
- Ingestion points: Handoff form in app/app/app.js.
- Boundary markers: Notes are clearly labeled and stored in a specific database field separate from agent instructions.
- Capability inventory: The skill has read/write access to Busabase bases.
- Sanitization: Notes are truncated to 2000 characters before storage in busabase-provider.js.
Audit Metadata