For AI agents

MCP for Agents

Vibecodr has an MCP endpoint built for AI agents at `https://openai.vibecodr.space/mcp`. Connect an MCP client to it and the agent can find public Vibecodr context, help you get a launch ready, and take supported signed-in actions — without ever getting the keys to your account.

Agent-facing MCP and workspace guidance for safe Vibecodr automation.

Implementation focus

Use this when you're connecting an MCP client or designing an agent workflow that should stay inside what Vibecodr safely lets agents do.

Expected outcomes

Connect agents to the right MCP endpoint, and know what Vibecodr will and won't share.
Choose agent work that fits — public discovery, launch prep, metadata help, and confirmed actions.
Know when to use the built-in MCP tools, when Code Mode applies, and where MCP stops.
Keep secrets, private drafts, storage internals, and deploy access out of agent work.

MCP Tools for Agents

Vibecodr's MCP surface exposes small, task-shaped tools at https://openai.vibecodr.space/mcp so an MCP client can help with Vibecodr work without inheriting raw platform authority.

Endpoint

Use the remote Streamable HTTP MCP endpoint at https://openai.vibecodr.space/mcp.

Auth boundary

Protected tools require Vibecodr OAuth/session checks. Upstream refresh tokens and platform grants stay server-side.

Product scope

MCP tools adapt Vibecodr product tasks. This is not an admin panel, widget surface, database shell, or unmediated platform control plane.

MCP client endpoint

https://openai.vibecodr.space/mcp

Use MCP For Agent Help, Not Admin Control

Connect an MCP client when you want an agent to understand Vibecodr context and help with supported product tasks. Public tools can read public Vibecodr context. Signed-in tools use your authorized session and stay limited to the actions Vibecodr exposes for agents.

MCP is a good fit for finding public examples, preparing a publish, checking runtime readiness, drafting share copy, or updating live metadata after confirmation. It is not a way to bypass the app, read private drafts, inspect secrets, operate storage directly, or receive deployment tokens.

Good fit: "Find public vibes like this and draft launch copy." Not a fit: "Give the agent raw database access or deployment tokens."

Useful Agent Workflows

Find public context

Ask an agent to look up public vibes, social previews, tags, share copy, and examples before you start a new idea.

Prepare a publish

Use agent help to review dependencies, import status, preview readiness, publish steps, and launch notes before anything goes live.

Polish after launch

Draft titles, descriptions, changelog notes, and follow-up ideas. Metadata updates still need the right account access and confirmation.

Keep private work private

Public MCP tools do not hand agents private drafts, secrets, raw logs, provider account identifiers, or storage internals.

Code Mode And Native Tools

Most MCP clients should use native Vibecodr tools. Each tool is shaped around a recognizable product task, so the agent can stay focused on the goal instead of driving a raw API catalog.

Code Mode is a separate, explicitly enabled workflow for generated code that calls a constrained host. Use it only when the MCP client or product flow offers it; otherwise, stay on the normal MCP endpoint.

Generated Code Mode code does not receive tokens, environment variables, or direct deployment wiring.
Default tool listings show product-shaped MCP tools.
Use https://openai.vibecodr.space/mcp for ordinary MCP clients.
If Code Mode is unavailable, switch back to native tools instead of asking the agent to guess around the boundary.

Where MCP Stops

MCP can help coordinate Vibecodr tasks, but it does not replace the product surfaces that own your work. Put browser UI in a vibe, trusted backend behavior in a Pulse, and scheduled or webhook-driven work in automations.

For backend integrations, use the Vibecodr capability that matches the job: env.fetch for policy-aware outbound calls, env.secrets for server-side credentials, and MCP tools for confirmed agent workflows.

Write Pulse handlers

Design safe automations

Understand public discovery

Understand Vibes and Pulses

What Vibecodr MCP tools are

Vibecodr MCP tools live at https://openai.vibecodr.space/mcp, and they're built for AI agents. Connect an MCP client when you want it to understand your Vibecodr context and help with supported product tasks.

Public tools can read public Vibecodr context. Signed-in tools use your authorized session, and they stick to the actions Vibecodr opens up to agents — nothing more.

If your agent runs on a server with no browser — an SSH session or CI — it can't finish the OAuth sign-in this endpoint needs. Drive the hosted tools through the Vibecodr CLI instead: run `vibecodr status` or `vibecodr doctor` for the commands to use and the no-browser sign-in.

Use MCP for public discovery, launch guidance, share copy, metadata help, and controlled signed-in actions.
Don't use MCP to get around the app, read private drafts, peek at secrets, run storage directly, or grab deploy tokens.
The default surface is Streamable HTTP at /mcp, with OAuth-compatible sign-in for protected actions.
Code Mode is separate — it only kicks in when the MCP client or product flow turns it on.

Remote MCP endpoint text

https://openai.vibecodr.space/mcp

Useful agent workflows

MCP is a great fit for finding public examples, getting a publish ready, checking that a vibe will run, drafting share copy, and updating live metadata after you confirm it.

Keep every step visible to you. A good agent can tell you exactly what it did — whether it searched public context, prepped a publish, checked readiness, drafted copy, or asked you to confirm a change.

Have agents search public vibes, tags, social previews, and examples before starting something new.
Let an agent review dependencies, import status, preview readiness, publish steps, and launch notes before anything goes live.
Draft titles, descriptions, changelog notes, and follow-up ideas before applying any metadata change.
Public MCP tools never hand an agent your private drafts, secrets, raw logs, provider account IDs, or storage internals.

Where MCP stops

MCP can help coordinate work, but it doesn't replace the product surfaces that actually own your work. Put browser UI in a vibe, trusted backend work in a Pulse, and scheduled or webhook-driven work in an automation.

For backend integrations, reach for the tool that matches the job: `env.fetch` for policy-aware outbound calls, `env.secrets` for server-side credentials, and MCP tools for confirmed agent workflows.

Point ordinary MCP clients at https://openai.vibecodr.space/mcp.
Signed-in actions still depend on your account access and permissions.
Code Mode code never receives tokens, environment variables, or direct deploy wiring.
If a workflow needs secrets, storage writes, schedules, or provider calls, build it as a Pulse or automation — not browser code.

Example and read next

Example: an AI agent needs to publish and check Vibecodr work. Connect it to https://openai.vibecodr.space/mcp, lean on the built-in tools by default, and save Code Mode for when it's explicitly turned on.

Use these related pages when you need the next layer of guidance. They point to the most likely follow-up tasks, not every page that happens to touch the same system.

Read next: Vibecodr CLI
Read next: Vibes & Pulses
Read next: Pulse SDK & Handlers
Read next: Automation Safety
Read next: SEO & Discovery