sdd
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to execute shell commands for testing, linting, and task validation as part of its core development workflow.
- Evidence: In
phases/execute.md, the instructions mandate running test suites, type checkers, linters, and custom validation strings defined intasks.yaml. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local project environment and user input which could potentially contain malicious instructions intended to manipulate the agent's behavior during the spec or execution phases.
- Ingestion points: Reads project-level documentation (
CLAUDE.md, build files), user-provided feature proposals (spec/<slug>/proposal.md), and task arguments. - Boundary markers: The skill does not define explicit delimiters or instructions to ignore embedded prompts within the processed files.
- Capability inventory: The skill has the capability to execute arbitrary local commands (tests/validation) and write to the local file system (
spec/directory). - Sanitization: No explicit sanitization or validation of the content within the ingested files is performed before processing.
Audit Metadata