# Research

> Pre-coding research and planning mode. Use before starting any task, feature, or change. Enforces thorough codebase study, pattern research, and explicit approval before implementation.

- **Type:** Skill
- **Install:** `agentstack add skill-anbturki-claude-toolkit-research`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [anbturki](https://agentstack.voostack.com/s/anbturki)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [anbturki](https://github.com/anbturki)
- **Source:** https://github.com/anbturki/claude-toolkit/tree/main/skills/research

## Install

```sh
agentstack add skill-anbturki-claude-toolkit-research
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Research & Plan

Do NOT write code yet.

## Step 1 — Understand

- Read the project instructions, architecture guidelines, and coding standards
- Study the existing codebase — understand current implementations, patterns, and structure
- Research real-world patterns and proven solutions (use MCPs and official docs when available)
- Verify all information before recommending — do NOT guess, predict, or assume

## Step 2 — Plan

- Ask yourself: How is this done elsewhere in the codebase? What can I reuse? Am I adding unnecessary complexity?
- Propose 2-3 approaches with trade-offs for each
- Reference specific files and patterns from the existing codebase
- Flag potential breaking changes, assumptions, and ambiguities
- Write the plan in a markdown file the user can review
- Frontend and backend should have separate plans if applicable

## Step 3 — Wait

- Present your proposed approach with reasoning
- List any assumptions or trade-offs
- Wait for explicit approval before implementing anything

## Constraints

- Do NOT over-engineer or predict future needs
- Do NOT propose creative/novel solutions — follow established practices
- Do NOT jump straight to coding for anything non-trivial
- Do NOT propose rewrites when a small change will do
- Do NOT plan more than one feature at a time

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [anbturki](https://github.com/anbturki)
- **Source:** [anbturki/claude-toolkit](https://github.com/anbturki/claude-toolkit)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-anbturki-claude-toolkit-research
- Seller: https://agentstack.voostack.com/s/anbturki
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.