# connect0

> Where your AI workforce runs — the AI workforce platform. connect0 gives every AI agent what a new hire gets on day one: an identity, a budget, a memory, and the business systems to do real work; every action audited, every dollar approved before it is spent, over OAuth 2.1 + MCP. Pay per use.

## When to use connect0

Reach for connect0 when the job is one of these:

- Give an AI assistant (Claude Code, Claude Desktop, Cursor, VS Code) governed access to company data and tools — connect via MCP at https://mcp.connect0.ai/mcp (OAuth 2.1; see /auth.md).
- Run agents on a schedule or from triggers with per-run audit trails, spend caps, and credit metering.
- Connect sources (GitHub, Postgres, and more) read-only into an isolated project workspace agents can work against.
- Manage a mixed team of humans and agents with scoped, default-closed access control.

Not a fit for: unmetered raw model API access, or hosting arbitrary web apps.

How to call: MCP (preferred for assistants) at https://mcp.connect0.ai/mcp, or REST at https://api.connect0.ai (OpenAPI: https://api.connect0.ai/openapi.json). Both use the same OAuth 2.1 bearer tokens — the handshake is documented at /auth.md.

## Getting started

- [Getting started (skill.md)](/skill.md): agent front-door manual — what connect0 is, how to authenticate, and the step-by-step getting-started workflow.

## Product

- [connect0 Agents — any agent, on your business](/agents/index.md): Any MCP-capable agent joins your business with its own identity, endpoints and workspace; the per-client setup picker is below the fold.
- [connect0 Controls — every dollar approved before it's spent](/controls/index.md): Budgets, caps, a published price before every action, audit, and your own keys; the rate table is the proof.
- [connect0 Blocks — business software your agents can operate](/blocks/index.md): The ledger first: provenance on every posting, dry-run before commit, approval gates, one API and one MCP contract.
- [Open by design — every machine-readable surface](/open/index.md): Every machine-readable surface on connect0.ai with a live curl for each and what it returns.
- [Safety and compliance status](/safety/index.md): What is in place, what is in progress, responsible disclosure, and a control-by-control status table with evidence.

## Pricing & company

- [Pricing — plans, credit packs, per-unit rates](/pricing/index.md): every number from the live credit catalog; 1 credit = $0.01.
- [How pricing works](/docs/pricing/index.md): the full billing model — credits, hourly rollup, subscriptions vs packs vs the starter grant, platform key vs BYOK, spend controls.
- [Docs index](/docs/index.md): the full help-center table of contents, grouped by product area.
- [Contact](/contact/index.md): how to reach the team (hi@connect0.ai; privacy@ and security@ for those topics).

## Assistant setup guides

Per-client instructions for connecting an assistant to the connect0 MCP server:

- [Claude Code](/agents/claude-code/index.md): Anthropic's terminal-native coding agent for Claude.
- [Claude Desktop](/agents/claude-desktop/index.md): Native macOS / Windows app for chatting with Claude.
- [Claude (web)](/agents/claude-web/index.md): Claude in the browser — chat, projects, custom connectors.
- [ChatGPT](/agents/chatgpt/index.md): OpenAI's assistant with custom connectors and a marketplace.
- [Cursor](/agents/cursor/index.md): AI-native code editor built on VS Code.
- [VS Code](/agents/vscode-copilot/index.md): GitHub Copilot Chat in VS Code, with MCP servers.
- [Windsurf](/agents/windsurf/index.md): Agentic IDE that pilots itself across your codebase.
- [Cline](/agents/cline/index.md): Autonomous coding agent that runs in your editor.
- [Zed](/agents/zed/index.md): Multiplayer-native code editor written in Rust.
- [JetBrains AI Assistant](/agents/jetbrains-ai/index.md): Built-in AI for every JetBrains IDE — including Junie.
- [Goose](/agents/goose/index.md): Block's open-source desktop and CLI MCP agent.
- [Continue](/agents/continue/index.md): Open-source AI code assistant for VS Code and JetBrains.
- [MCPJam](/agents/mcpjam/index.md): Postman-style inspector and playground for MCP servers.
- [Microsoft 365 Copilot](/agents/m365-copilot/index.md): Microsoft's enterprise AI across Office and Teams.
- [PostHog Code](/agents/posthog-code/index.md): PostHog's in-product AI for analytics and product engineering.
- [Postman](/agents/postman/index.md): API platform with native MCP server support and inspector.
- [LibreChat](/agents/librechat/index.md): Self-hosted multi-model chat with MCP tools and workflows.
- [Open WebUI](/agents/open-webui/index.md): Self-hosted UI for Ollama and OpenAI-compatible APIs.
- [Tome](/agents/tome/index.md): MCP-native local-first chat client for Mac, Linux, Windows.
- [Other / manual](/agents/other/index.md): Any MCP-capable client — point it at our endpoint.

## Developer resources

- [connect0 REST API — OpenAPI 3.1 spec](https://api.connect0.ai/openapi.json): every operation with typed schemas, operationIds, and the shared error model.
- [connect0 REST API — interactive reference](https://api.connect0.ai/docs): human-readable Scalar rendering of the same spec.
- [connect0 API catalog (RFC 9727)](https://api.connect0.ai/.well-known/api-catalog): machine-readable linkset of the API surface.
- [connect0 MCP server](https://mcp.connect0.ai/mcp): Streamable HTTP MCP endpoint (OAuth 2.1; server card at https://connect0.ai/.well-known/mcp/server-card.json).
- [connect0 authentication guide](/auth.md): OAuth 2.1 handshake for MCP and REST, token lifetimes, revocation.

## Docs

- [About connectors](/docs/about-connectors/index.md): What a connector is in connect0 — a project-scoped, credential-handled link to an external source or service that agents read and act through.
- [About skills](/docs/about-skills/index.md): What a skill is in connect0 — a reusable, versioned capability from the skill catalog that attaches to an agent, first-party or your own.
- [Access model](/docs/access-model/index.md): How connect0 decides who can do what — principals (humans and agents), teams, roles, project and resource scoping, and why agents are default-closed.
- [Accounts and projects](/docs/accounts-and-projects/index.md): How connect0 tenancy works — an account is your organization, a project is an isolated workspace, and some things scope to one, some to the other.
- [Agents Command Center](/docs/agents-command-center/index.md): The per-project agent roster at /a/<account>/p/<project>/agents — a fleet KPI strip plus a sortable, searchable, shareable table.
- [API and MCP reference](/docs/api-and-mcp-reference/index.md): Where the full REST API and MCP tool reference live, the connect0 service hostnames, and a quick-reference table of the main MCP tools.
- [Audit trail](/docs/audit/index.md): What connect0 records, where to view it — the account audit log at /settings/audit and the per-run audit trail on the execution view — and the event kinds you'll see.
- [Authoring a skill](/docs/authoring-a-skill/index.md): Write a skill in connect0 — the SKILL.md frontmatter and body, the six body kinds, and where in the dashboard to create one.
- [Bring your own key (BYOK)](/docs/bring-your-own-key/index.md): Add your own provider API key so model calls bill against your provider contract, with connect0 charging only a small orchestration fee.
- [Chatting with Zero](/docs/chatting-with-zero/index.md): Open Ask Zero from the sidebar, ask questions in plain language, and review, approve, or deny the action cards Zero proposes.
- [Choosing a model](/docs/choosing-a-model/index.md): How the model catalog works and how to pick, pin, and govern the runtime model an agent uses — balancing capability against cost.
- [Common issues](/docs/common-issues/index.md): Fast problem-cause-fix answers for the things that most often stop connect0 working — out of credits, expired connector auth, model or provider errors, and agents that produce nothing.
- [Connect a data source](/docs/connect-a-source/index.md): Register a connector (GitHub, Postgres, and more) so it materialises read-only into your project workspace for agents to read.
- [Connect your assistant](/docs/connect-your-assistant/index.md): Point an external MCP assistant — Claude Code, Claude Desktop, Cursor, VS Code — at connect0's MCP server and complete the one-time OAuth handshake.
- [Core objects](/docs/core-objects/index.md): A 60-second map of the connect0 object model — account, project, agent, thread, run — and where connectors, skills, and models fit.
- [Create an agent](/docs/create-an-agent/index.md): Create an agent in a project, give it a persona, model, skills, and connections, then invoke it interactively, on a schedule, or from a trigger.
- [Credits and pricing](/docs/credits-and-pricing/index.md): How connect0 meters usage against a credit wallet — what consumes credits, how balances work, and what happens when you run out.
- [Data and privacy](/docs/data-and-privacy/index.md): Where your data lives, how connection credentials and provider keys are encrypted, and how agent memory is scoped and protected.
- [Getting started (for agents)](/docs/getting-started/index.md): The agent-facing operating manual — MCP auth, what to connect, the full scoped tool catalogue, Ask Zero, credits, and the rules agents follow.
- [Managing billing](/docs/managing-billing/index.md): Buy credits and packs, manage your subscription, update your payment method, read invoices, and understand pending (unsettled) payments.
- [Managing skills](/docs/managing-skills/index.md): Enable, scope, version, and attach skills in a project — including importing from GitHub or npm and controlling which skills agents can reach.
- [Memory](/docs/memory/index.md): What agents remember across conversations — memory scopes (account, user, agent), how memories are retrieved into a run, and how you review them.
- [Model dispatch](/docs/model-dispatch/index.md): How a catalog slug becomes the model actually called at runtime — resolution, allow/deny, BYOK, and deprecation substitution.
- [Personas](/docs/personas/index.md): A persona is a reusable, account-scoped profile for how an agent communicates — tone and voice — separate from what it does or what it knows.
- [Quickstart — connect your first agent](/docs/quickstart/index.md): What connect0 is and the first three steps — create an account and project, then point an MCP agent like Claude Code at your core.
- [Running agents](/docs/running-agents/index.md): Dispatch an agent interactively or from a trigger, watch it execute live, and read run status and history.
- [Spend controls](/docs/spend-controls/index.md): Set a monthly spend cap and balance alert, and understand what happens when your account runs out of credits.
- [Team and principals](/docs/team-and-principals/index.md): Members of a connect0 account can be humans or agents. Both are principals; agents are default-closed, can't own the account, and can't grant access.
- [The model catalog](/docs/model-catalog/index.md): A reference for the connect0 model catalog — how entries are structured, the modalities and capabilities they carry, and how they are priced.
- [The public MCP server](/docs/public-mcp-server/index.md): Connect an external MCP assistant to connect0 at mcp.connect0.ai — the OAuth 2.1 handshake and the tool surface it exposes.
- [Threads and runs](/docs/threads-and-runs/index.md): A conversation with an agent is a thread; each agent turn is a run that produces messages. How to read a run's status, cost, and outcome.
- [Usage and metering](/docs/usage-and-metering/index.md): How connect0 measures what you use, rolls it up into credit burn each hour, and where to read your usage in the dashboard.
- [What is an agent?](/docs/about-agents/index.md): An agent is a configured, autonomous principal that runs on your account — a persona, a model, skills, and connections, with every run audited.
- [What is Zero?](/docs/about-zero/index.md): Zero is connect0's built-in, account-scoped assistant — it answers questions, inspects your account with read tools, and proposes actions you confirm before it acts.
- [Why did my agent run fail?](/docs/why-did-my-run-fail/index.md): The common reasons an agent run stops early — no runtime model, missing platform key, empty credit wallet, connector auth — and how to fix each.
- [Zero and your data](/docs/zero-and-your-data/index.md): How Zero grounds its answers and actions in your connectors, skills, memory, and the docs corpus — what it can touch, what it can't, and how every action is audited.

## Pages

- [About connect0](/about/index.md): The story behind connect0 — why we stopped bolting agents onto human-era software and built an AI-native operating core where data, tools, and agents connect under one governed surface.
- [Privacy notice](/privacy/index.md): What connect0.ai inc stores about you, why, who we share it with, and how to remove it.
- [Terms of Service](/terms/index.md): Terms covering your use of the connect0 service operated by connect0.ai inc.

## Blog

- [Leaving should be easy](/blog/leaving-should-be-easy/index.md): Openness is a claim anyone can type into a website. The kind worth anything is the kind you can check with a command. Here is the difference between assertable and verifiable openness, why we chose the second, and every command that proves it.
- [The connective core for AI agents](/blog/the-connective-core-for-ai-agents/index.md): As agents multiply, the bottleneck isn't the model or the chat — it's the connective core underneath. One governed place where every tool and data source connects, any agent runs, and everything is audited. Not a smart assistant beside your stack — the connective tissue under it.
- [Why MCP is the right shape for an agent backend](/blog/why-mcp-for-agent-backends/index.md): MCP is the standard that lets any agent plug into one connective core — decoupling capability from the substrate underneath, so you connect once and every agent runs on the same governed foundation.

## Guides

- [Connect a CSV to your agent](/guides/upload-a-csv/index.md): Connect a spreadsheet into the core — generate a pre-signed link, drop your file in a browser, and your agent reads it via `sh`.
- [Connect a data source to the core](/guides/connect-a-data-source/index.md): Plug a GitHub repo or a Postgres database into connect0 so your agent can read it — one connector, materialised into the project workspace.
- [Connect a dataset once, iterate on it forever](/guides/upload-then-iterate/index.md): Connect a file into the core once, then have your agent run analyses against it repeatedly. Use `info` to keep the agent oriented across turns.
- [Connect a dataset, analyze it, get results back](/guides/upload-analyze-download/index.md): The full round-trip on connect0 today — connect a dataset into the core, have your agent run analyses via `sh`, retrieve the results, and where each step has friction in the current toolset.
- [Connect a file straight from the dashboard](/guides/dashboard-drag-drop/index.md): Pure-browser upload path — no MCP, no terminal, no agent in the loop. For when a human is sitting at connect0.ai and just wants a file in the core the agents run on.
- [Connect a screenshot for your agent to analyze](/guides/upload-screenshot/index.md): Connect an image into the core — browser drag-drop, the agent reads it via `sh`, and the failure modes when content-type doesn't match.
- [Connect a video, get a transcript back](/guides/upload-video-for-transcription/index.md): Connect multi-hundred-MB media into the core — the curl flow, and what your agent does with the bytes once they're in.
- [Connect a whole directory tree as a tarball](/guides/upload-tarball/index.md): Connect a multi-file archive into the core and have your agent unpack it via `sh`. A worked example with the exact path-canonicalisation rules.
- [Connect big binaries over a slow network](/guides/big-binary-uploads/index.md): Connect multi-MB payloads into the core, why we bind `content_length` into the URL, and how to read a "size mismatch" 403 from R2.
- [Connect CI build artifacts into the core automatically](/guides/ci-auto-upload/index.md): Use a service token to connect build outputs into a connect0 project, so a teammate's agent can inspect them without a manual handoff.
- [Connect long context as a file, not a paste](/guides/long-context-as-file/index.md): Stop pasting 10,000-line logs into chat. Connect them into the core and have your agent grep, summarise, and ask follow-up questions against the file.
- [Egress + uploads together](/guides/egress-and-uploads/index.md): A project's `egress` flag controls outbound network from the sandbox — not inbound uploads. Here's the distinction and a few corner cases.
- [Get binary and large artifacts out of the core](/guides/download-workaround/index.md): Pulling a big binary out of the core today — chunked base64 through `sh`. Token-expensive and fragile, but it works until `download.create_link` lands.
- [Get results out of the core](/guides/download-today/index.md): Data out of the core is still Connect. What works today for pulling files back from `/workspace` — small text via `sh "cat"` — what doesn't, and where we're heading.
- [Govern what your agents can reach](/guides/govern-your-agents/index.md): Three governance guarantees that ship today in connect0 — default-closed agents with ACL enforced on the MCP connector path, an audit trail on every action, and spend caps that stop a run before it overspends.
- [Path canonicalisation rules — the core keeps writes in bounds](/guides/path-canonicalisation/index.md): What target_path you can pass to `upload.create_link` and `upload.write`. Why the rules are strict — no project can escape its own prefix — and how to work around them.
- [Rotate or revoke R2 keys — controlled credentials, one place](/guides/rotate-r2-keys/index.md): Rotate the access key and every outstanding pre-signed URL dies at once. The clean, operator-controlled rotation procedure across apps/cf-sandbox, apps/mcp, apps/web.
- [Run a code review in an isolated sandbox](/guides/code-review-sandbox/index.md): Drop a branch onto your connect0 core and let an agent run your test suite via `sh` in an isolated sandbox, reporting failures back — connect0 as a private CI runner your agents drive.
- [Run an agent that iterates on a config file](/guides/iterate-config/index.md): Hand a config to an agent running on your connect0 core; it edits in place via `sh` and verifies with `cat`. The subtle-but-powerful pattern that turns the sandbox into a live config editor.
- [Run an agent that keeps a persistent notebook on the core](/guides/agent-notebook/index.md): Let an agent running on your connect0 core keep `/workspace/notes/` as a persistent shared journal — reading, editing and appending across conversations.
- [Run an agent that pushes its results to GitHub](/guides/git-push-from-sandbox/index.md): Let an agent running on your connect0 core push its code-shaped output straight to GitHub with `sh "git push"` — the cleanest way off the sandbox until `download.create_link` ships.
- [Run an agent that scaffolds a new project](/guides/scaffold-from-template/index.md): Let an agent running on your connect0 core write boilerplate into a fresh `/workspace/` with `upload.write`. A worked example of an agent as code generator on the core.
- [Run Claude Code on your connect0 core](/guides/connect-claude-code/index.md): Three commands to point Claude Code at your connect0 core and let it run shell commands on your connected data, in an isolated per-project sandbox.
- [Run one agent across your repo and your database](/guides/one-agent-many-tools/index.md): One agent, two connected sources. Mount a GitHub repo and a Postgres database into the same project workspace, then let any MCP agent read both with `sh` to reconcile schema drift in a single run.
- [Scoped access — who can upload in a multi-user project](/guides/multi-user-uploads/index.md): Every member's write is scoped by role and recorded by name. Here's how the access model governs uploads, what `viewer` can't do, and conventions for keeping members from clobbering each other.
- [The agent's simplest write — `sh "cat > file"` for one-liners](/guides/sh-cat-redirect/index.md): The simplest way for an agent running on the core to write a file — pipe text into it via the shell. When this beats the dedicated upload tools, and when to stop reaching for it.
- [The upload audit trail — who wrote what, and when](/guides/upload-audit/index.md): Every upload-link issuance and every claim lands in `upload_link` — a governed, forensic record of who touched the workspace. Read the trail from the dashboard, from `info`, and (for operators) directly from Postgres.
- [Tiny configs, no curl — `upload.write`](/guides/upload-write/index.md): When the agent already has the bytes in hand, skip the pre-signed dance. `upload.write` PUTs inline content up to 1 MiB directly.
- [Upload links don't survive expiry](/guides/upload-link-expiry/index.md): Pre-signed PUT URLs are short-lived by design. Here's why, how long you get, and what to do when "the link expired" happens.
- [What `recent_uploads.status` means](/guides/upload-status/index.md): The pending → claimed → expired / oversize state machine on every upload link. Sourced from R2 events, not the client.
- [What an upload link can and can't do](/guides/upload-link-capability/index.md): A precise inventory of what a pre-signed `upload.create_link` URL grants — and what's deliberately out of scope.
- [When an upload silently fails](/guides/debug-silent-uploads/index.md): The upload completed, the dashboard says `pending` forever. Walk the queue path — Worker logs, HMAC drift, R2 event rule misconfig.
- [Why pre-signed URLs and not a proxy](/guides/why-presigned/index.md): The design decision behind the upload surface — zero connect0 bandwidth in the data path, capability bound by SigV4 + expiry. The cost / latency / trust tradeoffs.
- [Workspace storage + R2 costs, metered per project](/guides/workspace-costs/index.md): How much you can stash in `/workspace`, what Cloudflare charges for it, and how the core meters storage per project so the bill is always traceable to a source.

## Docs (full corpus)

- [/llms-full.txt](/llms-full.txt): Every page concatenated as a single markdown file.