sapiom-agent-builder
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The deployment documentation includes examples using
execSyncto package code into ZIP files. - [REMOTE_CODE_EXECUTION]: The skill enables deploying JavaScript code to
blaxel.services.sapiom.aifor scheduled execution. - [EXTERNAL_DOWNLOADS]: Local testing instructions involve downloading packages via
npm install. - [PROMPT_INJECTION]: The templates exhibit an indirect injection surface. Ingestion points: Scraped markdown in
templates/llm-agent.js. Boundary markers: Template system prompts. Capability inventory: Network requests and code deployment. Sanitization: Truncation is used but no content validation.
Audit Metadata