platonic-workflow

Pass

Audited by Gen Agent Trust Hub on Mar 8, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a coordinated four-phase development process. No patterns of prompt injection, role-play bypass, or safety guideline overrides were detected.
  • [SAFE]: File system interactions are restricted to local documentation directories (docs/drafts/, docs/specs/, docs/impl/) and the project codebase. No network operations, credential exposures, or sensitive file access (e.g., SSH keys, AWS config) occur.
  • [SAFE]: The skill manages internal dependencies within the vendor ecosystem (platonic-specs, platonic-impl, platonic-code-review) for code generation and review. These are well-scoped and do not involve remote code execution from untrusted sources.
  • [SAFE]: The skill has an indirect prompt injection surface (Category 8) because it processes local documentation to influence code generation. (1) Ingestion points: docs/drafts/, docs/specs/, and the codebase. (2) Boundary markers: Not explicitly defined. (3) Capability inventory: Code generation and file-writing via internal skills. (4) Sanitization: Not explicitly present in the workflow. This surface is standard for local development tools and is mitigated by the structured process and user confirmation gates.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 8, 2026, 11:32 PM