neon-theme
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: Accesses the sensitive credential file at
~/.build-host/credentials.json. This access is necessary for the skill's intended function of authenticating with thebuild.hostAPI to perform deployments. The risk is minimized as the data is sent only to the service associated with the credentials. - [PROMPT_INJECTION]: The skill processes untrusted input by reading 'current UI and constraints' from the project directory. This content is then used to generate code changes and trigger deployments.
- Ingestion points: Local project files and repository content referenced in SKILL.md.
- Boundary markers: None identified; the skill does not use delimiters to isolate processed code from instructions.
- Capability inventory: The skill modifies local repository files and performs network POST requests to the
build.hostAPI. - Sanitization: No sanitization or validation of the ingested UI code is described in the skill instructions.
Audit Metadata