Spec-driven development
Specify, decompose, and ship — with your AI agent.
Athenode turns one product spec into a detailed mind map, breaks it down into leaf specs, and hands each one to an AI agent to implement — all driven from the CLI you already use.
npx @athenode/cli initFeatures
From first question to shipped feature
Give your AI agents a clear spec instead of a vague prompt, and get exactly what you meant.
Clarify the spec
Answer a few questions, get a clear spec
AI asks the right questions and turns your answers into a clear spec.
- The agent asks what's missing, so nothing is left to guesswork
- Every answer lands in a living mind map
Plan the build order
Always know what to build next
Specs are linked by their blockers, so agents only pick up work that's ready to start.
- Blockers between specs set the build order
- Blocked specs wait until what they depend on is done
Ship with AI
Ship every piece with the agent you trust
Your AI agents build them in parallel while you track every result.
- Works with Claude Code, Cursor, Codex and more
- Every change is recorded against its spec
Decomposition in depth
Start with an idea. Go as deep as you need.
Capture the big picture first, then break each spec into smaller ones, level by level, until every leaf is small enough for an agent to implement in one run.
- Referral rewards program
- Referral link generation
- Unique invite codesReady
- Share buttons on the profile pageReady
- Reward payout rules
- Qualifying purchase checks
- Minimum order amountReady
- Refund clawback windowReady
- Crediting the account balance
- Ledger entry per rewardReady
- Reward notification emailReady
- Referral stats dashboard
- Invites and conversions chartReady
- Export to CSVReady
Integrations
Works with 9 AI coding tools
Athenode exports your agents, skills, rules and MCP servers to 9 AI coding tools, from Claude Code, Cursor and Codex to GitHub Copilot, Kiro and more — straight from the CLI, no manual setup.
See what each tool gets- Claude Code
Anthropic's agentic coding tool that reads your codebase, edits files and runs commands from your terminal, IDE or browser.
- Cursor
An AI-first code editor with agents that understand your whole codebase and plan, write and refactor code alongside you.
- Codex
OpenAI's coding agent that works in your terminal, your IDE and the cloud to write, review and ship code.
- GitHub Copilot
GitHub's AI pair programmer, with coding agents in VS Code, JetBrains IDEs, the Copilot CLI and on GitHub itself.
- OpenCode
An open-source AI coding agent for the terminal, desktop and IDE that works with the model provider of your choice.
- Google Antigravity
Google's agent-first development platform, where agents plan, build and verify work across the editor, terminal and browser.
- Devin
Cognition's autonomous AI software engineer, which takes on tasks end to end, from planning the change to opening the pull request.
- JetBrains Junie
JetBrains' coding agent for its IDEs and the command line, which plans and carries out multi-step tasks in your project.
- Kiro
An agentic IDE and CLI that turns prompts into clear specs, then plans and implements them with agents and hooks.
Your setup
Your setup, your rules
Every project gets its own agent setup — the agents, skills, MCP servers and rules your AI tools work with. Start from ours, make it yours, share it with the team.
Start from the default
Each project gets its own copy of the Athenode setup, ready from day one.
Make it yours
Clone it, then add or remove agents, skills, MCP servers and rules.
Share it with your team
Everyone on the project works with the same setup, kept in one place.
Autonomous execution
Your spec tree keeps moving
Athenode understands how specs depend on one another, then continues through the next implementation-ready leaf — autonomously.
Athenode completed 3 specs overnight while tracking dependencies automatically.
Pricing
Start free. Scale when your team does.
Every plan runs the full spec-driven workflow. Upgrade for unlimited specs, more agent setups and your whole team working from one spec tree.
Free
Try spec-driven development on your own projects.
Start for free$0
What's included
- Up to 3 free projects
- 100 specifications per project
- 1 agent setup per project
- 10 agents, 5 skills and 3 MCP servers per setup
Solo
For developers shipping with AI agents every day.
Coming soon
Everything in Free, plus
- Unlimited specifications
- 5 agent setups per project
- 30 agents, 30 skills and 15 MCP servers per setup
- 10 CLI tokens per project
- 5× higher CLI rate limit
- Recommended
Team
For teams planning and building from one spec tree.
Coming soon
Everything in Solo, plus
- Up to 10 project members
- Invites with editor and viewer roles
- 10 agent setups per project
- 50 agents, 50 skills and 30 MCP servers per setup
- 10× higher CLI rate limit
Business
For organisations scaling AI development across teams.
Coming soon
Everything in Team, plus
- Up to 50 project members
- Unlimited agent setups
- Unlimited agents, skills and MCP servers
- 20 CLI tokens per member
- 20× higher CLI rate limit
- Priority support
Included in every plan
- Spec tree with mind-mapping and decomposition
- AI agents implement leaf specs end to end
- Export to Claude Code, Codex, Cursor and more
- Real-time progress tracking for every spec
Compare all plan limits
| Feature | Free | Solo | Team | Business |
|---|---|---|---|---|
| Free projects per user | 3 | Not applicable | Not applicable | Not applicable |
| Project members (incl. owner) | 1 (owner only) | 1 (owner only) | 10 | 50 |
| Invites, editor/viewer roles | No | No | Yes | Yes |
| Specifications per project | 100 | Unlimited | Unlimited | Unlimited |
| Agent setups per project | 1 | 5 | 10 | Unlimited |
| Skills per setup | 5 | 30 | 50 | Unlimited |
| Agents per setup | 10 | 30 | 50 | Unlimited |
| MCP servers per setup | 3 | 15 | 30 | Unlimited |
| CLI tokens per user per project | 1 | 10 | 10 | 20 |
| API requests per minute per project (CLI) | 60 | 300 | 600 | 1200 |
| Priority support | No | No | No | Yes |
FAQ
Common questions
An agent setup is the configuration of your AI agents, skills, MCP servers and AGENTS.md for a project — created and customised by anyone on the team, then shared with every project member so everyone works from the same configuration.
No. The CLI writes its content into a block between Athenode markers, and everything outside that block stays exactly as you wrote it. MCP servers are merged into your existing configs rather than replacing them. If the markers are broken, the CLI stops with an error and changes nothing — and --dry-run shows the full plan before anything is written.
Every member has one of three roles. Viewers get read-only access, editors can change specs and agent setups, and owners can also invite, re-role and remove members. You invite people by email — there's no separate accept step, and they get access the first time they log in.
No. Mind-map, decompose, implement is a starting point, not a mandate. The /athenode-init skill looks through your repository and recommends settings for your gitflow, testing, code review and methodology. Your own rules go in each settings file's Custom section, which later runs never overwrite, and you can clone the default setup and change anything in it.
No. Your code stays in your repository, and your AI tool works on it locally. Athenode stores your specs and agent setups, plus, for each implemented spec, the paths of the files that changed and a short summary of what was done.