What happens if a subagent provides an incomplete report?

The skill includes a mandate to reject and redeploy subagents if they report conclusions without verifiable sources or specific code locations.

Is the plan compatible with multiple chat sessions?

Yes, the plans are structured to be self-contained with their own references, making them ideal for execution across different chat contexts or session boundaries.

How does this skill prevent AI hallucinations?

It enforces a mandatory Phase 0 documentation discovery step where the AI must cite specific sources, file paths, and method signatures before any implementation planning begins.

What is the Orchestrator and Subagent model?

The Orchestrator manages high-level synthesis and plan authoring, while Subagents are deployed specifically for fact gathering and data extraction to ensure technical accuracy.

Can I use this for existing private codebases?

Yes, the skill is designed to use grep and file reads to discover existing patterns and documentation within your local repository to maintain architectural consistency.

Structured Implementation Planner

Name: Structured Implementation Planner
Author: thedotmack

bythedotmack

•

66,391

•

生产力与工作流

Generates phased, documentation-first implementation plans to ensure high-accuracy code execution without API hallucinations.

The Make Plan skill transforms Claude into a rigorous Orchestrator that breaks down complex features into manageable, evidence-based phases. It prioritizes a mandatory Documentation Discovery phase where subagents verify actual API signatures and patterns before any code is written, effectively preventing common AI hallucinations. By enforcing a strict reporting contract and verification checklists for every step, it ensures that multi-part coding tasks remain grounded in existing project patterns and official documentation across long-running development sessions.

主要功能

01Mandatory Phase 0 Documentation Discovery to verify APIs before implementation

02Strict subagent reporting contracts requiring source citations and exact file paths

03Self-contained session boundaries allowing plans to be executed consecutively in new contexts

04Phased implementation structure with built-in verification checklists and anti-pattern guards

05Orchestrator-Subagent delegation model for rigorous fact-checking and synthesis

0666,391 GitHub stars

使用场景

01Planning large-scale feature migrations where API accuracy and pattern consistency are critical

02Managing complex, multi-step coding tasks that require structured oversight and verification

03Integrating unfamiliar libraries or frameworks by first enforcing a thorough documentation crawl

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add thedotmack/claude-mem make-plan

For use in Claude.ai and ChatGPT