Issues, code reviews, documents, chat and call transcripts are indexed into one base and answered under your own permissions. Ask the in-app assistant, or give Claude Code or Cursor the same base over MCP, from issue to deploy.
Code4 is in early access: start a 14-day free trial, or join a small first cohort of design partners.code4.space is developed, reviewed, built and deployed on code4.space.
Indexed from
One knowledge base
A keyword pass and a vector pass, merged and expanded along links, answered under your permissions.
Asked from
This is the center of Code4: one index over seven sources, answered under the permissions you already have, for your team in the app and for your agents over MCP. Nothing here is a separate product to buy or a wiki to keep alive.
Seven sources are indexed: chat, issues, code reviews, project documents, call transcripts, commit messages and the code of each default branch. Code on other branches is not indexed; branches come back only as links hanging off a result that already matched.
A question runs a keyword pass and a vector pass at once, the two rankings are merged, and the result is expanded along the links between an issue, the review that closed it and the commits that named it. Take the embedding service away and the vector pass drops out, while the keyword pass keeps answering.
Ask the assistant inside the app, or connect the base over MCP: streamable HTTP, sign-in through Keycloak, and every call runs as you rather than as a service account. The same endpoint carries tools for issues, reviews, branches, pipelines and deploys, so a coding agent reads the base and works in your tracker without a second account.
11 custom field types, parent and subtask trees, checklists, watchers, saved filters and bulk edits, with an estimate and a running timer on the issue itself. Boards add columns with WIP limits, sprints, swimlanes and a backlog panel, and cards move by mouse or by keyboard.
Merge requests with per-line threads, comments on a selected range of the diff, suggested edits that land as a commit, reviewer approvals, and a marker showing how much of the diff you have already read.
Documents live in a folder tree with version history and access rules per node. Chat, threads and video messages sit next to calls with screen sharing and recording. After a call you get a transcript, a summary, chapters and action items with owners, once participants consent.
Git over HTTPS is served by the platform itself: clone, push, branches, tags, file browsing, blame and branch comparison. Protection rules live beside the repository instead of in a separate admin tool.
Steps declare their inputs and outputs, so a run is a graph rather than a list, and a step can run hermetically with a digest-pinned image and an outbound allowlist. You write the pipeline in Java, Go or TypeScript against a typed SDK, and when a step fails you can open a shell in the workspace it left behind. If you would rather not write code, the same SDK compiles a .ci/pipeline.yaml into the same model.
Environments, deploys and rollbacks. A deploy can wait for a named person to approve it, and deploys to your own Kubernetes go through an agent inside the cluster, so your kubeconfig never leaves it.
An assistant on every screen and an MCP server with more than 150 tools, so Claude Code or any MCP client can read the knowledge base and work with your issues, reviews, pipelines and docs. The in-app assistant shows every tool call and asks before risky actions. An agent you connect over MCP acts with the permissions of the person who signed it in. The branch protection rules you set hold for that agent as they do for that person. Required approvals apply to merge requests, so restrict push to a branch to make a merge request the only way into it.
Push a repository, open issues, write documents, talk in chat, hold a call. None of it is a knowledge base chore: there is no article editor here and no wiki to keep current.
Each of the seven sources is split into fragments and indexed as it is written. A call joins the base only if the organization left transcript indexing on and nobody in the room opted out, and it leaves again when the retention window ends.
One pass covers chat, issues, reviews, documents and transcripts, over exactly what you are allowed to open. The answer names the source it came from, so a decision can be traced back to the thread where it was made.
The base is an MCP endpoint. An agent signs in as you, reads it, and, when you let it, opens issues, comments on a review or starts a pipeline. Nothing it can reach is anything you could not reach yourself.
Code4 works best for 2 to 20 engineers on a new or young project, where one person can decide how the team works.
You start client projects often. Give each client one knowledge base, so every decision, review and call stays findable next to the repository, board and pipeline, instead of wiring five tools every time.
You are setting up engineering for the first time. Start with one knowledge base that your team and agents share from the first issue, instead of migrating off a patchwork later.
You are a few engineers who already write code with Claude Code, Cursor or Copilot. Give your agent the knowledge base, along with your issues, reviews and pipelines, through one MCP connection.
We are opening Code4 to a small first cohort of teams. You run one real project on one knowledge base that your team and your AI agents share, we work with you directly, and what you need shapes what we build next.
You can leave at any time without giving a reason. Your code is in git, so a mirror keeps a full copy at any time, and we export your issues and documents on request.
We reply to every application within 2 business days.
Code4 runs hosted today. Your deploys can still land on your own infrastructure.
We run Code4 for you. Sign up and start with a 14-day free trial, then continue on a paid plan.
Deploy to your own Kubernetes. An agent inside your cluster pulls deploy jobs, so credentials never leave it.
Need Code4 itself on your own infrastructure? Talk to us about your requirements.
Two hosted plans billed per user each month, and enterprise terms on request. Team and Business start with a 14-day free trial, no card required. After the trial, the subscription is billed monthly by card through Stripe. Teams in the design partner program use Code4 at no charge during the program and get a founding-partner discount when they move to a paid plan.
Hosted on code4.space, for a team that wants the whole loop in one place.
12 US dollars per user per month
Hosted on code4.space, for an organization that needs answers faster.
24 US dollars per user per month
For larger organizations, custom terms and on-premises needs.
Contact us
No, and there is nowhere to write it. The product has no article editor and no wiki section: the index is built from what your team writes anyway, which is chat, issues, code reviews, project documents, call transcripts, commit messages and the code on your default branches. What you decide is whether call transcripts and source code join it at all, and any participant can keep a call out.
One database and one permission model. A question about a decision searches issues, review threads, documents, chat and call transcripts in a single pass, because they are not in four products. Pipelines are typed code in a language you already write, and a YAML file compiles into the same model if you prefer one.
A document or a folder can be closed to everyone except the people you name, and the creator stays on that list. Search runs the same permission check the screens do, and it is built the other way round from most: the filter is assembled before the query and an empty set of permissions returns nothing, so a private document does not surface as a hint to someone who cannot open it.
It answers questions in a chat panel, it comments on merge requests as AI Review, it explains why a pipeline step failed when you ask, it drafts an outline of a code review once an admin turns that on, it writes a summary and a transcript for a call, and it powers search across the organization. It does not autocomplete in an editor. The in-app assistant asks before risky actions, and an agent you connect over MCP acts with the permissions of the person who signed it in.
Only what the AI features touch. The assistant, call summaries and issue drafts send their input to Anthropic, and so do AI code review and explanations of failed pipelines, which are on by default and which an admin can turn off. Speech from calls, voice and video messages goes to Deepgram for transcription, and search uses an OpenAI embedding model, which also receives commit messages and the source code of each repository's default branch: sending that code is on by default, and an admin can turn it off. Everything else stays in our database and our file storage, as the Privacy Policy says. If the embedding service is unavailable, search keeps answering from the keyword pass alone.
Repositories move by pushing them: the remote is an ordinary git server over HTTPS, so git push --mirror is the whole job. Issues and documents have no importer yet, and until one exists the REST API and the MCP endpoint are the way in. They are the same interfaces the app itself uses, so nothing is held back from them. During the design partner program we move your first issues and set up your first pipeline with you, by hand.
Code4 runs hosted today. Your deploys can target your own Kubernetes cluster through an agent that runs inside it, so your kubeconfig never leaves the cluster. If you need Code4 itself on your own hardware, talk to us.
Four known gaps. No importers from other trackers. Documents autosave with a lock rather than letting two people type at once. Call transcription runs in one language, set in configuration rather than picked per call, and there is no way to upload a recording that did not happen here. And it is a web app: layouts collapse for a narrow window, but there is no public mobile app yet.
A small first cohort of teams runs one real project on Code4 at no charge for the length of the program, with one knowledge base that the team and its AI agents share. We help with setup and put your requests first. In return we ask for a short check-in, weekly in the first month and then every two weeks, and honest feedback.
You choose. Move to a plan on the pricing page with a founding-partner discount, or leave and take your data with you. Nothing is charged until you choose a plan.
No. Start with one new project on Code4 and keep everything else where it is. Code4 does not sync with GitHub today, so pick a project that can live in one place.
Your code is in git, so a mirror keeps a full copy at any time. Issues and documents are available through the API, and we export them on request. When a paid subscription ends, our terms give you 30 days to ask for an export.
An agent you connect over MCP works with the permissions of the person who signed it in, so give it the access you are comfortable with. The branch protection rules you set hold for it as for that person. Required approvals apply to merge requests, so restrict push to a branch to make a merge request the only way into it. Protected environments wait for an approval before each deploy. The in-app assistant shows every tool call and asks before risky actions.
Yes. Code4 runs an MCP server: add it to your client, sign in, and your agent works with your permissions across issues, reviews, pipelines, docs and calls.
Tell us about your team and the project you would start with. A founder reads every application.
This opens an email in your mail app. Nothing is sent until you press Send.
Or write to support@code4.space. We reply to every application within 2 business days. If we record the fit call, we ask you first.