Fimo
For developers

Deliver the site.
Keep your focus.

You become the site’s support channel the day it goes live: every headline change or testimonial swap lands as a Slack ping mid-work. Fimo ends that. Deliver on your stack, hand over the keys, keep the review and drop the busywork.

One review queue
instead of Slack pings.

Agents propose. You merge.

The rest of the team edits the site on the preview and in Content while the code stays yours, in your repository. Every agent run lands in its own branch with a report, a preview, and a diff. Nothing reaches production until you merge it, and questions wait for you in the Inbox.

How autonomous agents work
your-repo
.fimo/agents/
seo-audit/
GOAL.mdREQUIRED
config.yamlREQUIRED
capabilities.yamlOPTIONAL
SOUL.mdOPTIONAL
scripts/OPTIONAL
references/OPTIONAL
assets/OPTIONAL
content-translator/
post-deploy-check/
.envGITIGNORED

Hand over a site that improves itself.

Agents in your repo. Yours to shape.

Most handovers start slow decay: stale content, SEO that never happens, a redesign six months later. A Fimo site keeps agents working on it: files in the repo, versioned, scheduled, every run waiting for your review.

How custom agents work

Every change is a branch. Every review is one diff.

Isolated environments. One diff to review.

Agents don’t experiment in production. Every branch spins up a disposable copy of your stack in milliseconds. When the work is done, you read one diff covering code, schema, content, and labels together.

How isolated environments work
New post ready for review
FieldValue
TitleHow agentic workflows actually finish
Slugstructuring-agentic-workflows
Entry IDae873e4a-af94-4d1b-ae46-4a80f10747bb
Publish date2026-06-11

Deploy once.
Site and agents, live.

Global edge, SSL, CDN. No second pipeline.

One fimo deploy pushes your site and your agents to a preview URL, no second pipeline. Publish when the preview looks right. A bad release? Restore a previous version from history. You decide what goes live.

How deployment works
How it works

From handoff to autopilot.

Connect.

Paste the setup prompt into your coding agent. It runs npx fimo@latest init on the code you already have and wires your content so it becomes editable.

Delegate.

Marketing edits without you. Agents run on a schedule in isolated branches and leave a report, a preview, and a diff for your review.

Deploy.

fimo deploy gives every branch a preview URL. Publish when it is right, and restore a previous version from history if a release goes wrong.

Built by the team behind Strapi.

Fimo comes from the founders of Strapi, the leading open-source headless CMS. A decade of content infrastructure for millions of developers, distilled into a platform built for the agent era.

70K+
GitHub stars
Millions
of developers served
FAQ

Questions developers ask.

No. Build with Claude Code, Codex, Cursor, or whatever you already use, on your usual stack: Next.js, Astro, React, Vite, TypeScript. Fimo is framework-agnostic: it works over the output and turns it into an editable, autonomous site without changing your workflow. Details on your AI stack.

Yes. Fimo is a git-based, framework-agnostic CMS: use it as a Next.js CMS, a React CMS, or an Astro CMS, and it works the same with Vite, React Router, or SvelteKit. It runs over whatever your framework outputs, so nothing about your build changes; your content simply becomes editable, and every edit lands as a clean commit in your repo.

Your coding agent runs npx fimo@latest init --project-id in your repo. It installs the Fimo SDK for your framework and wires your content pieces so they become editable, without touching the rest. No proprietary layer between you and your codebase.

Yes. The site stays real code in a repository you control. Every action relies on Git and is reversible.

It breaks its own environment, not your site. Each agent works in a sandbox with a dedicated server, database, and media library, and nothing reaches production until the branch is merged. Delete the branch and the experiment never happened.

Yes. Every run works in its own branch and ends with a report, a preview, and a diff. You merge it, or you close the branch and the experiment never happened. When an agent needs a decision mid-run, it asks in the Inbox and waits for you.

Yes. An agent is a folder in your repo, .fimo/agents/<name>/, with a GOAL.md for the mission and a config.yaml for the triggers, schedule, and model. Scaffold one with fimo agents create, or prompt your coding agent (Claude Code, Cursor, whatever you use): Fimo ships a skill for it, so you describe what you want and it writes the agent for you.

None run by default. The CLI ships templates you scaffold in one command: an SEO audit, a post-deploy check, a content translator, and a broken-link sweeper. Run the ones that fit, edit their goals, or build your own from scratch.

From Fimo Studio. Text and images directly on the preview in edit mode, everything else as structured fields in Content, and the chat for whatever goes beyond content. Every edit is versioned.

It is a CMS built the way developers work: Git underneath, real code, isolated environments, reviewable diffs, and a content layer on top so the rest of the team never needs you for a headline again.

Start with your website.

Try Fimo online in seconds, then connect your coding agent and publish when you are ready.

Try Fimo for free