airweave
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python helper script
scripts/search.pyto facilitate API requests and format search results. - [EXTERNAL_DOWNLOADS]: The search script connects to the official Airweave API at
api.airweave.aito retrieve data from user-connected applications. - [PROMPT_INJECTION]: The skill ingests data from external applications which may contain adversarial instructions, presenting an indirect prompt injection surface.
- Ingestion points: Search results containing messages and document content retrieved from the Airweave API via
scripts/search.py. - Boundary markers: The output uses Markdown headers but does not include specific boundary delimiters or instructions to isolate external content.
- Capability inventory: The skill allows the execution of the
search.pyscript via the command line. - Sanitization: No explicit validation or filtering is performed on the retrieved data before it is presented to the agent.
Audit Metadata