spec-to-plan
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is entirely instructional, providing a template and a step-by-step workflow for documentation and planning tasks. It does not include any executable code or configuration that would grant elevated permissions.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data (PRDs). While this presents a surface for indirect prompt injection, the risk is mitigated by a mandatory human-in-the-loop step where the agent is instructed to quiz the user for approval before generating the final output.
- Ingestion points: External Product Requirement Documents (PRDs) or specifications provided by the user in Step 1.
- Boundary markers: None explicitly defined in the Markdown template to separate instructions from data.
- Capability inventory: The skill uses codebase exploration tools and file system write operations to the
./plans/directory. - Sanitization: No specific sanitization or filtering of input data is defined.
- [COMMAND_EXECUTION]: While the skill mentions exploring the codebase, this is a standard behavior for agent skills and does not involve the execution of arbitrary or malicious commands. All file system activity is confined to reading context and writing plan documentation.
Audit Metadata