blog-from-markdown
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFECREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill instructs the agent to establish authentication by reading a sensitive file at
~/.build-host/credentials.json. Accessing local filesystem credentials is a sensitive operation, though expected for the skill's deployment purpose. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its processing of external data. Ingestion points: Markdown folders provided by the user (SKILL.md). Boundary markers: None present to distinguish user content from agent instructions. Capability inventory: The skill can execute local checks and perform network operations via API endpoints (POST /api/projects, POST /api/projects/:uuid/deploy) as documented in SKILL.md. Sanitization: No validation or sanitization of the user-provided markdown content is specified before generation or deployment.
Audit Metadata