vibe
Warn
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local shell script at ~/.agents/skills/here-now/scripts/publish.sh to handle the deployment of generated files during Phase 3 and Phase 4.
- [EXTERNAL_DOWNLOADS]: If the deployment script is missing, the skill prompts the user to download and install a third-party package from an unverified source (heredotnow/skill) using npx.
- [REMOTE_CODE_EXECUTION]: The skill dynamically generates code based on untrusted user input. 1. Ingestion points: User requests describing site content. 2. Boundary markers: None identified. 3. Capability inventory: Subprocess call to publish.sh. 4. Sanitization: No evidence of escaping or validation for user input before it is written to index.html or app.js.
Audit Metadata