MCP
PlungeAI MCP Server Quickstart
Connect any MCP client to `https://mcp.plungeai.com/v1`.
The PlungeAI MCP server gives any MCP client the 20 plungeai_* tools: discover capabilities, run tools, agents, workflows and missions, schedule jobs and read results.
When to use PlungeAI MCP?
All four surfaces use the same ozk_ key, the same account and the same data.
| Surface | What it is | Use it when |
|---|---|---|
MCP server, https://mcp.plungeai.com/v1 | The 20 plungeai_* tools over MCP | An AI client operates PlungeAI live: discover, run, save, schedule, remember, with approvals and continuations |
One API, https://api.plungeai.com | REST routes, OpenAPI at GET /v1/openapi.json | You write or generate code that calls PlungeAI |
Skills, https://skills.plungeai.com | 31 instruction packs your AI client loads | Your client should already know PlungeAI's conventions; skills complement a connection |
Ocean Studio, https://studio.plungeai.com | The browser app | A person builds visually, connects accounts and reviews runs |
A docs MCP server that searches and reads these pages from your agent is planned, tracked as TI-35: see Docs MCP.
Endpoint
| Item | Value |
|---|---|
| URL | https://mcp.plungeai.com/v1 |
| Transport | Streamable HTTP, JSON-RPC 2.0 over POST |
| Authentication | Authorization: Bearer $PLUNGE_API_KEY or X-API-Key: $PLUNGE_API_KEY (an ozk_ key); every request needs it, tools/list included |
| Responses | JSON; tools/call streams as server-sent events |
Create the key at Dashboard → One API → Keys (https://dashboard.plungeai.com/one-api?tab=keys (opens in a new tab)) and export it as PLUNGE_API_KEY before you configure a client.
Quick installation
Let your agent install it
Paste this prompt into your coding agent; it reads the setup file and connects itself:
Use curl to read https://docs.plungeai.com/agents.md and perform the setup to connect PlungeAIOr name the server directly:
Add the MCP server https://mcp.plungeai.com/v1 named plungeai to my user-level client config, with the header Authorization: Bearer and the value of my PLUNGE_API_KEY environment variable, then call plungeai_whoami to confirm the connection.Client installation
Claude Code, one command:
claude mcp add --transport http plungeai https://mcp.plungeai.com/v1 \
--header "Authorization: Bearer $PLUNGE_API_KEY" --scope userCodex, which reads the key from PLUNGEAI_API_KEY:
export PLUNGEAI_API_KEY="$PLUNGE_API_KEY"
codex mcp add plungeai --url https://mcp.plungeai.com/v1 --bearer-token-env-var PLUNGEAI_API_KEYA project .mcp.json that reads the key from each developer's environment, safe to commit:
{
"mcpServers": {
"plungeai": {
"type": "http",
"url": "https://mcp.plungeai.com/v1",
"headers": { "Authorization": "Bearer ${PLUNGE_API_KEY}" }
}
}
}One-click installs open your client's install dialog with the URL filled in; add the key afterwards in the client's user-level config:
Opens Cursor's MCP install dialog.
Opens VS Code's MCP install dialog.
Adds the server to Replit Agent.
Step-by-step pages per client:
Verify in any client: ask it to use plungeai_whoami to confirm your identity.
These docs use PLUNGE_API_KEY; the PlungeAI skills, the Claude Code plugin and the install page use PLUNGEAI_API_KEY; any name works if your config and shell agree.