Docs: start here.
The reference documentation and architecture receipts are being published on codemonster.dev. Until then, this page routes you to what is live on codemonster.ai: setup guides, the operations overview, environments, primitives, and the security hub.
The full reference publishes on codemonster.dev. Everything linked below is live today; when the reference publishes, these cards link out.
Want the guided version? Learn walks the same path end to end — a setup guide per MCP client and blueprint walkthroughs. This page is the map: what is documented, and where it lives.
Install. Connect. Build.
Three pages, in order, each ending where the next begins. All three are published today; the desktop builds themselves are still being finalized, so step 01 ends in an access request rather than an installer.
- 01
Install
CodeMonster is invite only while we onboard early adopters. Request an invite and we scope what you want to build, set the desktop app up with you, and stay with you to the first shipped solution.
Request an invite - 02
Connect your agent
Paste the MCP config block into Claude Code, restart it, and run whoami. The guide names what 401, 403 and "Tenant context required" mean and what to do about each.
Claude Code setup guide - 03
First build
Copy the field-service inspection blueprint prompt, adjust the vocabulary, and let the agent build on governed rails: begin-solution-edit, preview-solution-edit, commit-solution-edit.
Open the blueprint
Where each thing is documented.
Overview pages on codemonster.ai now; the per-tool reference and the architecture receipts move to codemonster.dev when it publishes.
Install, sign in, first five calls
The desktop app, the license login, and the calls every session starts with: whoami, list-tenants, use-tenant, get-domain-knowledge, begin-solution-edit.
Start with the desktop appThe control surface: approximately 100 named operations
Every operation by category, with what it stages and what it commits. The full per-tool reference publishes on codemonster.dev when live.
Operations overviewDev, staging, production; promote and roll back
The custom-frontend chain (develop → staging → production, owner / editor / viewer) and the platform publish lifecycle (Current → Testing → Production), explained separately.
EnvironmentsObjects, forms, workflows, grids, screens
What platform apps are made of, and how a prompt becomes create-object, add-field-to-form, create-workflow, generate-screenflow.
PrimitivesComponents, auth flow, hosting
What talks to what and how every request is authorized. The architecture receipts publish on codemonster.dev; the summary is live on the security hub now.
Architecture summaryA documented REST API over your real tables
Every object is a real Postgres table, and the platform REST API is documented with Swagger and served next to the API itself, so the reference always matches the running version. Your systems read, write and export the same data the app uses, under the same authentication and permissions. Not linked here: you reach it from inside the platform once you are signed in, not from this site.
Feeds and indexes.
The monster has many eyes. Some of them are crawlers. These are the files they read.
- /llms.txt Every public route with a one-line description, for agents that read sites.
- /rss.xml Blog posts and changelog entries in one feed.
- /sitemap-index.xml The sitemap, drafts and placeholders excluded.
When the docs are not enough.
- Support. Write to support@codemonster.ai or use the contact form. Include the operation you called and the exact error text.
- FAQ. Every question, one page — pricing, bring your own AI, building, environments, security, ownership.
- Community. Where builders gather — office hours, the showcase, and what shipped this week.
- Security. Vulnerability reports go to the disclosure page, not the forum.
Next: connect your agent.
The Claude Code setup guide runs from the config block to your first committed solution.