Hamster vs Spec Kit
Spec Kit gives your coding agent a fixed order to work in, from spec to code, and it’s very good at that. Hamster gives you a Brief and a Plan your agent builds from, and it still works when your PM, your designer and a second engineer all have a say.
Free for 10 Briefs a month, and viewers are always free.
The short version
Spec Kit gives you a fixed order of commands for your coding agent, from specify to converge. Every step writes markdown into your repo, and it’s free and MIT licensed.
Hamster gives you a Brief, a Plan with acceptance criteria on every task, and decisions that stay on the record, synced into your repo for whichever agent you run. It works when you build alone, and it keeps working when your PM, your designer and a second engineer have a say.
You can also run both. Hamster takes the speckit commands in as a Method your whole team shares.
What Spec Kit is good at
Spec Kit gives your coding agent a fixed order to work in. You write a constitution once, and then every feature goes through the same commands, from specify and clarify to plan, tasks, implement and converge. Each step writes markdown into your repo, so the spec, the plan and the task list sit next to the code and show up in every diff. The commands install as agent skills, and the docs list forty supported agents, including Claude Code, Cursor, Codex and Copilot. It’s MIT licensed and free.
The speckit commands, from the Spec Kit docs. github.github.io, September 18, 2026
Where it gets hard with a team
Say your team is adding SSO sign-in for the enterprise plan. Your PM wants it, your designer owns the sign-in screens, and two engineers will build it across the web app and the auth service.
One engineer runs /speckit.specify, and spec.md lands in a numbered folder on their branch. Your PM and your designer can read it in a pull request and leave comments, and every edit goes back through that engineer. Spec Kit’s maintainers are upfront that it doesn’t mandate how you collaborate, and they suggest a shared branch and pull requests on the spec. Meanwhile the second engineer needs a spec for the auth service, and two branches reaching for the next folder number is a known collision.
So nobody really owns the spec. Later, /speckit.converge checks the code against spec.md and adds tasks for whatever it missed. What your PM agreed on a call isn’t in that file, so converge can’t see it. The agent is checking the code against a spec your team never actually agreed on.
What Hamster does instead
Your team agrees on one Brief
The Brief lives in your Hamster workspace instead of on someone’s branch. Your PM describes the change in chat, and Hamster drafts the Brief from the tickets, the code and the call transcript behind it. Your designer and both engineers edit that same Brief and suggest changes. When it’s right, each of them marks it ready, and you can see who has and who hasn’t.
Every agent builds from one Plan
Hamster turns the agreed Brief into one Plan, with tasks across both services and acceptance criteria on every task. The tasks sync to Linear or Jira, so your tracker and the Plan are the same list. Each engineer runs hamster sync, and the Brief and the Plan land in .hamster/ in their clone. Claude Code, Cursor or Codex picks them up from there.
Review checks what you agreed
When the work comes back, you review the pull request in Hamster against the acceptance criteria in the Brief. You’re checking it against what your team signed off, so nobody has to remember what was said on the call.
Keep using Spec Kit
Nothing you’ve built gets thrown away. The speckit commands are agent skills, and Hamster keeps skills in a library your team shares. The setup prompt pushes them into your workspace, where you group them under a Method, so everyone runs the same steps with the same wording.
Your spec files stay in the repo, and your agent runs /speckit.plan and the rest exactly as it does today. The only change is that it runs them against the Brief and the Plan you agreed on, which hamster sync puts in a .hamster/ folder next to your specs.
Set up Hamster and bring Spec Kit along
Paste this into Claude Code, Cursor or Codex in your repo. You sign in once in the browser, and your agent does the rest.
Set up Hamster in this repo: follow https://tryhamster.com/plugin/install, then run hamster init and hamster sync. Then install Spec Kit (`uv tool install specify-cli`, then `specify init . --integration <your agent>`), copy the speckit-* skills it generated under .agents/skills/, and run `hamster skills sync --push-only` so they appear in my Hamster workspace. Stop when you need me to sign in.
Spec Kit is free. Why pay for Hamster?
Spec Kit is free and open source under the MIT license, and you pay only for your coding agent.
Hamster is free too, for 10 Briefs a month, and viewers are always free. After that it’s $40 per creator a month, so you pay for the people who write Briefs and not for everyone who reads them.
The $40 pays for the part a file can’t do. Hamster drafts the Brief from your tickets, your code and your calls, and keeps what you decided for the next one. When a teammate joins, they edit the same Brief and mark it ready, and the sync gets the result to every agent.
| What is compared | Spec Kit | Hamster |
|---|---|---|
| Price | Free, MIT licensed. You pay for your coding agent. | Free for 10 Briefs a month with viewers free, then $40 per creator a month. |
| Account | None. It runs from your repo. | You sign in once in the browser. |
| Where the spec lives | Markdown on the branch you ran it from. | A Brief in your workspace, synced into every clone. |
| Who signs it off | Whoever merges the pull request on the spec. | Each teammate marks the Brief ready. |
Questions
- Can we use Spec Kit and Hamster together?
- Yes. Your agent keeps running the speckit commands from your repo, and the specs you already have stay where they are. Hamster supplies the Brief and the Plan those commands run against.
- I work alone. Is Hamster for me?
- Yes. Hamster drafts the Brief from your code and your tickets, and the Plan gives your agent tasks with acceptance criteria. What you decide stays on the record for the next Brief, a Cloud Agent can work the Plan for you, and it’s free for 10 Briefs a month.
- We’re three people. Is this overkill?
- No. Hamster exists for the moment three people read one ticket three different ways, and small teams hit it fast because everyone ships. Viewers are free, so the people who only read a Brief and mark it ready don’t cost you anything.
- Does Hamster run the /speckit commands, or does my agent?
- No, your agent still runs them, from your repo, exactly as it does now. Hamster gives those commands an agreed Brief and Plan to run against. If you’d rather hand the work off, a Cloud Agent can work the Plan and open the pull request.
- Does it work with Claude Code, Cursor and Codex at the same time?
- Yes. hamster sync writes the Brief and the Plan into .hamster/ as markdown, and any agent that reads your repo can use them. There’s an MCP server too. One of you can be on Cursor and another on Claude Code, and you both build from the same Plan.
Agree on the spec before anyone’s agent builds it
Free for 10 Briefs a month, and viewers are always free. You set it up with one prompt in your repo.
More reading: spec-driven development for a team, BMAD Method vs Hamster and Hamster for Claude Code.
Sources, checked September 15, 2026: Spec Kit command reference, supported agents, license, discussion 2230, discussion 2116. Seen something out of date? Tell us at hello@tryhamster.com.