deep-plan

Create research.md and plan.md artifacts before gated code implementation.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/escotilha/claude-public --skill deep-plan-escotilha
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-plan
Source: https://github.com/escotilha/claude-public/tree/main/skills/deep-plan
Command: npx skills add https://github.com/escotilha/claude-public --skill deep-plan-escotilha

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prevents risky, premature coding by forcing a disciplined workflow where the AI researches and writes a detailed plan first, then waits for your explicit approval before implementing anything.

Core Features & Use Cases

  • Three-phase workflow with persistent artifacts: produces research.md and plan.md (plus .deep-plan-state.json) so work can be resumed and audited across sessions.
  • Hard-gated implementation: never writes code until the user approves the written plan.
  • Deep codebase understanding: performs deep reads (with targeted file discovery) to capture architecture, data flow, constraints, and open questions before planning.

Quick Start

Use the deep-plan skill to plan and (after you approve) implement a feature by giving a clear request like: “/deep-plan Refatorar sistema de cache”.

Frequently Asked Questions about deep-plan

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

FAQPage Schema
How do I prevent AI from writing code prematurely in a complex engineering workflow?

A research-plan-implement workflow enforces a hard approval gate before coding begins. It generates persistent markdown artifacts for research and planning, ensuring the AI never writes code until you explicitly approve the plan.

How does a gated approval process handle resumable engineering tasks across sessions?

Resumable engineering tasks are handled by tracking execution state via a JSON file and producing persistent markdown artifacts. This allows you to pause complex engineering workflows and resume them across different sessions without losing context.

What is the best way to plan complex codebase refactoring before implementation?

The best way to plan complex refactoring is a three-phase workflow that produces scoped markdown artifacts. It forces deep codebase reading to capture architecture and data flow, followed by writing a detailed plan that awaits your final approval before implementation.

Does the deep-plan workflow require writing a YAML frontmatter operational plan?

Yes, the deep-plan workflow requires a YAML frontmatter-defined operational plan. This structured plan ensures explicit user gating for final approval and proper state tracking throughout the research, planning, and implementation phases.

When should I use a hard-gated research and planning workflow for software engineering?

Use a hard-gated research and planning workflow for complex engineering tasks requiring deep codebase understanding. It is essential when you need scoped artifacts, open question tracking, and strict execution control to prevent risky, premature coding.