executing-plans

Executes written implementation plans by delegating tasks to Codex with review checkpoints.

Updated Jul 15, 2026
One-click install
npx skills add https://github.com/em411/supercodex --skill executing-plans-em411
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: executing-plans
Source: https://github.com/em411/supercodex/tree/main/skills/executing-plans
Command: npx skills add https://github.com/em411/supercodex --skill executing-plans-em411

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning a written implementation plan into working code requires disciplined task-by-task execution, verification, and review; this Skill orchestrates that process so plans are followed exactly without skipped steps or silent failures. ## Core Features & Use Cases - Critical plan review before execution: Loads the plan, raises concerns with the user before starting, and creates todos for each task. - Codex-delegated implementation: Classifies each task, picks the appropriate GPT-5.6 model tier, dispatches work to Codex, and reviews every diff against the plan with fix cycles via session resume. - Automatic fallback mode: If the Codex preflight fails, tasks are implemented inline instead, with the fallback announced explicitly. - Use Case: You have a multi-task implementation plan from a planning session and want it executed in a separate session with per-task verification, code review, and a clean handoff to branch completion. ## Quick Start Ask Claude to execute the implementation plan in plans/my-feature.md using the executing-plans skill.

Frequently Asked Questions about executing-plans

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I execute an implementation plan with Claude Code?

Invoke the executing-plans skill with a written plan file. It reviews the plan critically, creates todos, dispatches each task to Codex for implementation, reviews the diffs, runs verifications, and finishes the development branch when all tasks complete.

How does Codex delegation work during plan execution?

Each task is classified and routed to a GPT-5.6 model tier (sol, terra, or luna) per the codex-delegation skill, then dispatched using the codex-implementer-prompt template. Review findings are sent back to the same Codex session via --resume-last fix dispatches.

What happens if Codex is unavailable during execution?

The skill runs a Codex preflight once at the start. If it fails, fallback mode is announced and all tasks are implemented inline by Claude instead of being dispatched to Codex.

When should I use subagent-driven-development instead of executing-plans?

Use subagent-driven-development whenever subagents are available, since it is the primary supercodex execution path. Executing-plans is the inline alternative for sessions without subagent support.

When does plan execution stop and ask for help?

Execution stops immediately on blockers such as missing dependencies, failing tests, unclear instructions, critical plan gaps, or repeated verification failures. The skill asks for clarification rather than guessing.