Fimo
Build

Build custom agents.

Describe what it should do and pick when it runs. Fimo brings the skills, keeps the agent as code in your repo, and runs it next to your website. Every run lands in a branch for your review.

Under the hood

Technical by design.

Fimo brings the skills. The rest is plain code you version, deploy, and tune, down to each agent's model and triggers.

Skills are built in

Deploying, editing content and schema, SEO, translations, media generation. The Fimo CLI ships every skill and agents load them on demand with full project context. Your brief stays a few plain lines.

Agents are code

An agent is a folder of plain files: GOAL.md, config.yaml, and optional capabilities and scripts. Versioned, diffed, and reviewed like the rest of your codebase. No black box.

They run next to your site

Agents run on Fimo alongside your deployed site with access to its code, content, and assets. They act on schedules and events, not in a chat window.

A branch per run

Every run gets its own branch with an isolated sandbox and database. Review the report, the preview, and the diff, then merge. Nothing touches production on its own.

Any model, per agent

Give each agent its own model in config.yaml: Claude Opus 5, GPT-5.6 Luna, Gemini 3.7 Flash, Grok 4.6, DeepSeek V4 Pro, and more. A frontier model when precision matters, a fast one when it does not.

Runs keep their history

Every run keeps its report, its diff, and its logs in Studio. The next run starts with the project's content, assets, and previous runs as context.

Your workflow

They live in your repo.

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

Versioned and independent

An agent is a folder of plain files under .fimo/agents, with its own history in your repo. Each agent stays independent: edit one, deploy, and the others keep running as before.

Reviewed by anyone

Every run ends in a branch your team can review in Studio: devs, editors, PMs. Open the report, check the preview, read the diff, then merge or discard.

Any AI

Author with Claude Code, Codex, Cursor, or whatever your team runs. Fimo is attached to your site, not a coding tool.

Superpowers

Enhanced context. Real actions.

Every agent runs in its own branch with read and write access to the code, content, and assets that make your site. If you can describe it, an agent can propose it.

Generate reportsPropose changesDraft new entriesOptimize LighthouseTranslate releasesRefresh screenshotsGenerate reportsPropose changesDraft new entriesOptimize LighthouseTranslate releasesRefresh screenshots
Archive unused assetsA/B test headlinesFix broken linksAnything you can describeAudit accessibilityDraft changelogsSummarize analyticsArchive unused assetsA/B test headlinesFix broken linksAnything you can describeAudit accessibilityDraft changelogsSummarize analytics
Localize landing pagesTrack keyword rankingsTag new productsCompress hero videosDraft the newsletterSyndicate the blogLocalize landing pagesTrack keyword rankingsTag new productsCompress hero videosDraft the newsletterSyndicate the blog
Read and write access toCodebaseDatabaseContentAssetsRuns history
FAQ

Questions about agents.

Coding agents (Claude Code, Codex, Cursor) help your developers write code. Fimo agents run on your deployed site. They translate, audit, validate, and propose changes in a branch for the whole team to review.

Skills are the capabilities the Fimo CLI ships with: deploying, editing content and schema, SEO, translations, media generation. Agents load them on demand with full project context (content, assets, history, previous runs), so a few lines of brief are enough to get real work done.

On Fimo, next to your website. Each run gets its own branch with an isolated sandbox and database, works on the site code, content, and assets, and leaves a report, a preview, and a diff for review.

With your website. An agent is a folder in your repo, so fimo deploy ships the site and every agent definition together. Edit a goal or a schedule, deploy, and the next run uses it.

Yes, the way you roll back any file. Agents are versioned in your repo, so restore a previous GOAL.md or config.yaml, deploy, and the agent works the way it did before. Your other agents and your site are untouched.

Set model in each agent's config.yaml. GPT-5.6 Luna is the default. You can also pick Claude Opus 5, Claude Sonnet 5, Claude Fable 5, GPT-5.6 Terra or Sol, Gemini 3.7 Flash, Grok 4.6, DeepSeek V4 Pro, GLM-5.2, Kimi K3, or MiniMax M3. Run a frontier model where precision matters and a fast one where it does not.

Every run keeps its report, its diff, and its logs in Studio, and each new run starts with the project's content, assets, history, and previous runs as context. Nothing is hidden: what an agent knows is in your repo and in Studio.

No. Anyone can describe an agent in plain language in GOAL.md and pick when it runs in config.yaml. Your developers wire up anything custom in scripts, and anyone with review permission approves the work.

Agents read the branch you point them at, main by default. Every change they propose lands in their own branch with its own database, so production is never touched until you merge.

Yes. A cron schedule with its timezone, a webhook after every deploy, an event such as content being published, or only when you ask. Whatever cadence fits.

Every run is a proposal in its own branch. Review the report, the preview, and the diff, merge what is right, and discard the rest. Nothing reaches your site without a person merging it.

No. Whether they use Cursor, Codex, Claude Code, or anything else, Fimo agents work the same. They're attached to your site, not to a coding tool.

In AI credits, like the rest of Fimo: 1,000 credits are one dollar. Free includes one active schedule that runs at most once a day. Pro includes 20 active schedules down to a five-minute interval.
Keep exploring

Where to next.

Isolated environments

Every branch and every agent run gets its own sandbox with a dedicated server, database, and asset bucket. Changes are safe until they reach production.

Autonomous agents

Agents and your team working side by side. Every run lands in its own branch, so you review the diff and merge when it is right.

Start with your website.

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

Try Fimo for free