plan-to-code
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various git commands (
git status,git rev-parse,git add,git commit,git merge-base) to manage project state, pin base commits for future diffs, and finalize task implementations. These are standard operations for a development-focused tool.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon content from external files such as implementation plans, specifications, and Architecture Decision Records (ADRs), creating a surface for indirect instructions.\n - Ingestion points:
SKILL.md(Step 4) specifies reading the plan header, target tasks, specs, and cited ADRs from the local file system.\n - Boundary markers: The skill implements a restricted context-loading strategy in Step 4 and requires explicit user confirmation of test results in Step 7 before proceeding to completion steps.\n
- Capability inventory: The skill is capable of writing code (Step 5), editing the plan file (Step 9), and executing git commands to stage and commit changes (Step 10).\n
- Sanitization: No specific sanitization or escaping mechanisms are mentioned for the content read from external project artifacts.
Audit Metadata