implement-feature

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard Git commands (git pull, git worktree add) and filesystem operations (cd) to manage the development environment. These commands are localized to the repository and are standard for the described workflow.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes data from external, potentially untrusted sources (JIRA tickets and local PRD files).
  • Ingestion points: Data enters the agent context via JIRA URLs or local PRD file paths specified in the /implement-feature command (SKILL.md, Step 1).
  • Boundary markers: The instructions do not specify explicit delimiters or "ignore instructions" warnings when processing external requirements.
  • Capability inventory: The agent has the capability to execute shell commands (Git), write files (implementation_plan.md, task.md), and execute testing tools (playwright-cli, appium-mcp).
  • Sanitization: The risk is significantly mitigated by a mandatory "HARD STOP" in Step 1, which requires the user to review and approve the implementation_plan.md before any code implementation or tool execution begins.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 12:06 AM