opus-plan

Delegates planning tasks to a Claude Opus Plan subagent with ultrathink prompting.

1|Updated Feb 6, 2026
One-click install
npx skills add https://github.com/ETdoFresh/claude-marketplace --skill opus-plan-etdofresh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: opus-plan
Source: https://github.com/ETdoFresh/claude-marketplace/tree/main/plugins/opus-plan/skills/opus-plan
Command: npx skills add https://github.com/ETdoFresh/claude-marketplace --skill opus-plan-etdofresh

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Complex implementation tasks often need deeper reasoning than a quick in-context answer provides. This Skill offloads planning to a dedicated Claude Opus Plan subagent prompted with "ultrathink", producing thorough, file-grounded implementation plans saved to disk. ## Core Features & Use Cases - Deep Reasoning Delegation: Launches a Plan subagent running on the Opus model with an ultrathink prompt to maximize reasoning depth via prompting rather than the extended thinking API. - Context-Aware Planning: Passes the task description plus relevant conversation context (files, constraints, goals) so the subagent reads actual code before recommending changes. - Persistent Plan Output: Writes the full plan to a timestamped Markdown file under ./tmp for later reference. - Use Case: Before refactoring a large module, invoke /opus-plan to get a step-by-step implementation plan with file paths, risks, trade-offs, and open questions. ## Quick Start Invoke /opus-plan with a description of the implementation task you want planned, or ask for a deep implementation plan for your current problem.

Frequently Asked Questions about opus-plan

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

FAQPage Schema
How do I get a deep implementation plan for a coding task?

Invoke /opus-plan with your task description, or ask for a high-quality plan. The Skill launches a Claude Opus Plan subagent with an ultrathink prompt that reads relevant files and produces a step-by-step plan with risks and trade-offs.

Does opus-plan use the extended thinking API?

No. Reasoning effort is elicited through the ultrathink keyword in the prompt, not configured via budget tokens in the extended thinking API. The subagent runs on the Opus model with prompt-based depth instructions.

Where is the generated plan saved?

The full plan is written to ./tmp/claude-plan-<timestamp>.md, where the timestamp is generated with date +%Y%m%d-%H%M%S. The Skill creates the ./tmp directory if it does not exist and reports the file path to you.

What model does the planning subagent use?

The subagent is launched with subagent_type Plan and model opus, meaning it runs on Claude Opus. The prompt instructs it to read relevant files before recommending changes and to explore trade-offs and alternatives.

When should I not use opus-plan?

Avoid it for trivial tasks that need no deep reasoning, since spawning an Opus subagent adds latency and cost. It is best reserved for architecture reviews, complex refactors, and implementation strategies requiring thorough analysis.