feature-auto-pipeline

Pass

Audited by Gen Agent Trust Hub on Mar 19, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The orchestrator agent executes arbitrary bash commands for running test suites, linters, and type checks. It also performs Git operations (add, commit) on the local repository. The sub-agent executes shell commands to run TDD tests and uses Playwright for browser-based end-to-end testing.
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its document ingestion process.
  • Ingestion points: The orchestrator reads markdown files from specs/features/ (Requirements, Technical Design, and Task Plans) as the primary source of instructions for the development tasks.
  • Boundary markers: The skill uses basic markdown headers and placeholder interpolation ({{...}}) but lacks robust delimiters or 'ignore' instructions to prevent the sub-agent from following malicious commands embedded within the task descriptions.
  • Capability inventory: The sub-agent has capabilities for file system modification (writing code/tests), executing bash scripts, performing SQL queries via Supabase MCP, and controlling a browser via Playwright MCP.
  • Sanitization: There is no evidence of sanitization, filtering, or validation of the content extracted from the specification files before it is injected into the sub-agent's prompt template.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 19, 2026, 10:10 AM