recipe-front-build

Pass

Audited by Gen Agent Trust Hub on Sep 23, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from markdown task files located on the filesystem, which creates a surface for indirect prompt injection.
  • Ingestion points: The skill computes a "Consumed Task Set" by reading files matching the {plan-name}-task-*.md pattern from the docs/plans/tasks/ directory.
  • Boundary markers: While the skill references an "Orchestrator Execution Boundary" and a separate orchestration guide, it lacks explicit delimiters or instructions for the agent to ignore embedded prompts within the task files themselves.
  • Capability inventory: The skill possesses significant capabilities, including the ability to invoke subagents for code execution, perform integration tests, and execute git commit operations.
  • Sanitization: There is no evidence of content sanitization or validation for the task files before their contents are passed to the task-executor-frontend subagent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 23, 2026, 10:09 AM
Security Audit — agent-trust-hub — recipe-front-build