db-postgres
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates by inspecting the existing project stack, which involves reading untrusted local files. This creates an attack surface for indirect prompt injection where malicious instructions hidden within a project could influence the agent's actions during database setup or deployment.
- Ingestion points: Existing project files during the initial stack inspection phase (SKILL.md).
- Boundary markers: No specific delimiters or safety warnings are specified to isolate project content from the agent's instructions.
- Capability inventory: Local repository editing, environment variable modification via API, and project deployment via API (SKILL.md).
- Sanitization: No explicit sanitization or filtering of the ingested project data is defined.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill is configured to access sensitive authentication tokens stored at
~/.build-host/credentials.json. This access is standard for the vendor's toolset and is used to authenticate with the build.host API.
Audit Metadata