Skip to content
BA Toolkit

BA Toolkit

Turn a rough project idea into a complete, structured specification — requirements, user stories, acceptance criteria, API contracts, wireframes, and a step-by-step implementation plan.
Get started View on GitHub

Why BA Toolkit?

Section titled “Why BA Toolkit?”

29 interconnected skills

From /discovery (brainstorm an idea) through /brief, /srs, /stories, /ac, /nfr, /datadict, /apicontract, /wireframes, /scenarios, /handoff to /implement-plan — a complete BA pipeline inside your AI coding agent.

12 domain references

SaaS, Fintech, E-commerce, Healthcare, Logistics, On-demand, Social/Media, Real Estate, iGaming, EdTech, GovTech, AI/ML — each with domain-specific interview questions, typical entities, NFR categories, and glossary terms.

Zero runtime dependencies

The CLI is a single Node.js file. npm install -g '@kudusov.takhir/ba-toolkit' and you have the whole pipeline available offline. No build step, no config, no lock-in.

Works with 5 AI agents

Claude Code, Codex CLI, Gemini CLI, Cursor, and Windsurf. All five use native Agent Skills (SKILL.md) — no conversion, no .mdc, no plugins. Install once, use everywhere.

Industry-standard rigour

Every artifact follows established standards used by senior business analysts at top consultancies (BABOK, IEEE 830, ISO 25010, and others — see the glossary for details). Built-in traceability, provenance fields, and quality checks.

Notion + Confluence publish

ba-toolkit publish bundles every artifact into import-ready folders for Notion (Markdown) and Confluence (HTML). Drag-and-drop, no API tokens, no network.

How it works

Section titled “How it works”

Most skills follow the same cycle: Command → Context → Interview → Generate → Refine. Each skill loads all previous artifacts plus the domain reference, asks targeted questions (one at a time, table of options, one marked Recommended), writes a Markdown artifact, and offers refinement skills before moving on.

Every artifact links back to its predecessors — requirements trace to user stories, stories trace to acceptance criteria, and so on through the entire specification. Run /trace to verify nothing was lost and /analyze to catch quality issues before they become expensive rework.

After /handoff, run /implement-plan to produce a dependency-ordered implementation plan an AI coding agent can execute step by step — every task references the requirements and acceptance criteria it implements.

Quick install

Section titled “Quick install”
Terminal window
npx '@kudusov.takhir/ba-toolkit' init

Prompts for project name, domain, and AI agent, then creates AGENTS.md, output/, and installs the 29 skills into your agent’s directory.

Read the full getting-started guide →

Common questions

Section titled “Common questions”
Do I need all 29 skills? No. The lean pipeline covers essentials in ~3-4 hours. Add more steps when you need deeper coverage.
What if I don't know what to build? Start with /discovery — it runs a structured brainstorm and recommends a domain, name, and MVP scope.
Can I use it in any language? Yes. The agent detects your language and generates all artifacts accordingly.
How do I share with stakeholders? Run ba-toolkit publish to bundle artifacts for Notion or Confluence import.
What does the output look like? Browse a complete example project (Lumen Goods) with all 16 artifacts.

Have a question or found a bug?

Section titled “Have a question or found a bug?”

Open an issue on GitHub → — whether it is a feature request, a bug report, or a question about how to use a specific skill.