Research-Plan-Implement Skills

A set of agent skills that provide a structured workflow for codebase research, implementation planning, and plan execution.

Installation

Install the skills using the skills CLI with the repository’s url:

npx skills add https://github.com/MaibornWolff/acf-research-plan-implement-skills.git 
# or for SSH:
npx skills add [email protected]:MaibornWolff/acf-research-plan-implement-skills.git 

To install globally (available across all projects):

npx skills add https://github.com/MaibornWolff/acf-research-plan-implement-skills.git -g
# or for SSH:
npx skills add [email protected]:MaibornWolff/acf-research-plan-implement-skills.git -g

To install specific skills only:

npx skills add https://github.com/MaibornWolff/acf-research-plan-implement-skills.git --skill rpi-research
# or for SSH:
npx skills add [email protected]:MaibornWolff/acf-research-plan-implement-skills.git --skill rpi-research

Workflow

The three skills form a sequential pipeline. Each step produces an artifact that feeds into the next:

┌──────────────┐  ┌──────────────┐              ┌──────────────┐
│ rpi-research │─▶│   rpi-plan   │─────────────▶│rpi-implement │─▶ Working Code
└──────────────┘  └──────────────┘              └──────────────┘
  (optional)       Creates a phased              Executes the plan
  Investigates     implementation plan,          phase by phase with
  the codebase,    iterates with                 automated and manual
  produces a       the user                      verification
  written report

You can run all three steps end-to-end for a large feature, or start at any stage. Research can be skipped for simple codebases or straightforward features where the agent can plan directly without prior investigation.

Example: Adding rate limiting to an API

Here is how you would use the full pipeline to add rate limiting to an existing API.

Step 1: Research the codebase

Start by understanding how the API currently handles requests, what middleware exists, and where authentication is enforced. The research skill produces a report at docs/agents/research/.

/rpi-research How does request handling work in this API? I want to understand
the middleware chain, how routes are registered, and how authentication is enforced.

This creates a report like docs/agents/research/2026-04-02-api-request-handling.md that documents the current middleware pipeline, route registration patterns, and relevant file paths with line numbers.

Key points for good research prompts:

• Focus on understanding what exists today, not on the future feature
• Be specific about which areas to investigate (middleware, routes, auth)
• Mention the broader goal so the agent knows what’s relevant to cover

Step 2: Plan the implementation

Reference the research report so the planning skill can build on it directly instead of re-investigating the codebase.