developer-spec-workflow
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The workflow involves reading user-provided briefs and source materials to generate artifacts.\n
- Ingestion points: Raw brief and source material stored in
docs/brief.md.\n - Boundary markers: Intake evidence is separated into its own file, though no specific prompt delimiters are mentioned.\n
- Capability inventory: The agent creates implementation code and executes shell commands to verify the runnable sample.\n
- Sanitization: No specific sanitization of external brief content is required by the instructions.\n- [DYNAMIC_EXECUTION]: The workflow requires the agent to generate and then execute scripts to provide end-to-end verification.\n
- Evidence: Instructions direct the agent to 'Build the Runnable Proof' and execute documentation commands from a clean state.\n- [COMMAND_EXECUTION]: The skill requires executing shell commands for repository inspection and sample validation.\n- [SAFE]: The skill enforces secret hygiene by instructing the agent to keep credentials and private keys out of version control.\n- [SAFE]: References an external community resource for provenance without introducing executable dependencies.
Audit Metadata