skills/openai/plugins/executing-plans/Gen Agent Trust Hub

executing-plans

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • Process-Driven External File Ingestion: The skill defines a workflow to read and execute instructions from a plan file. This creates a potential surface for indirect prompt injection if the plan content originates from an untrusted source.
  • Ingestion points: The agent reads a plan file in Step 1.2 to create task lists.
  • Boundary markers: No specific technical delimiters are mentioned for the plan content, though the skill provides instructional boundaries.
  • Capability inventory: The skill facilitates task execution and verification, potentially involving file modifications or command execution via subagents.
  • Sanitization: The skill includes explicit instructions to 'review critically,' 'identify any questions or concerns,' and 'ask for clarification rather than guessing' before beginning implementation. These manual checkpoints serve as a primary safeguard against unexpected instructions within the plan.
  • Integration with Developer Tooling: The documentation recommends the use of external tools such as Claude Code, Codex CLI, Copilot CLI, and Gemini CLI. These are well-known technology services provided by established organizations used for software development tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 11:16 AM
Security Audit — agent-trust-hub — executing-plans