About Spaces by Peakflo
Spaces by Peakflo is a hosted MCP server for finance and operations teams to hand repetitive work to cloud agents or people. It also exposes workflow execution, read-only data queries, approval tasks, and connected app operations.
Key features
- 01Create one-off or recurring tasks
- 02View and run automated workflows
- 03Query data tables and analytics reports
- 04Search Business Context documents
- 05Handle approvals and call connected apps
Use cases
- 01Schedule recurring finance requests for AR ageing summaries
- 02Delegate vendor information collection
- 03Check for failed workflows
- 04Query unpaid bills by vendor
MCP support
Provides a remote MCP server using Streamable HTTP and OAuth 2.1 with PKCE. MCP clients can create tasks, run workflows, query data, and use connected apps.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://api.peakflo.ai/mcpclaude mcp add --transport http ai-peakflo-spaces https://api.peakflo.ai/mcp{"mcpServers":{"ai-peakflo-spaces":{"type":"http","url":"https://api.peakflo.ai/mcp"}}}Open in Cursor{"servers":{"ai-peakflo-spaces":{"type":"http","url":"https://api.peakflo.ai/mcp"}}}Open in VS Code{"mcpServers":{"ai-peakflo-spaces":{"type":"http","url":"https://api.peakflo.ai/mcp"}}}{"mcpServers":{"ai-peakflo-spaces":{"serverUrl":"https://api.peakflo.ai/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMESpaces by Peakflo: MCP server
Hand off repetitive work to the cloud. An agent or a human gets it done.
Spaces by Peakflo is a hosted (remote) Model Context Protocol server for finance and operations teams. Connect it to Claude, ChatGPT, Gemini, Cursor, VS Code or any MCP client, describe the job, and hand it off.
- Server URL:
https://api.peakflo.ai/mcp - Transport: Streamable HTTP
- Auth: OAuth 2.1 with PKCE (S256). Clients can register through dynamic client registration or a client ID metadata document. You sign up or sign in in the browser, so there are no API keys to paste.
- Website: https://spaces.peakflo.ai · Docs: https://spaces.peakflo.ai/docs/mcp/
This repository contains no server code. It only holds the connection configs for each client and the directory metadata.
Connect
| Client | How |
|---|---|
| Claude (claude.ai / Desktop) | Settings → Connectors → Add custom connector → https://api.peakflo.ai/mcp |
| Claude Code | claude mcp add --transport http peakflo https://api.peakflo.ai/mcp — or as a plugin: /plugin install peakflo-spaces --marketplace peakflo/spaces-mcp |
| ChatGPT | Settings → Apps → Developer mode → Create app → https://api.peakflo.ai/mcp |
| Gemini app | gemini.google.com → Settings → Connected Apps → Custom apps → Add a custom app (US, personal accounts) |
| Gemini CLI | gemini extensions install https://github.com/peakflo/spaces-mcp |
| Cursor | Install the plugin from this repo, or add mcp.json to ~/.cursor/mcp.json |
| VS Code | code --add-mcp '{"name":"peakflo","type":"http","url":"https://api.peakflo.ai/mcp"}' |
| Other | Any MCP client that supports Streamable HTTP and OAuth: use the URL above |
The first time you connect, a browser window opens. Sign up or sign in, pick your workspace and click Allow access.
What your assistant can do
| Tool | What it does | Kind |
|---|---|---|
task_create | Hand off a one-off or recurring (cron) task to a cloud agent or a person. Returns a link. Queued work is never shown as done. | write |
workflow_list, workflow_get, workflow_execution_list | See your automations and their runs | read-only |
workflow_execute | Run a workflow | write, destructive |
table_list, table_query, table_sql | Query data tables (read-only SQL) | read-only |
report_schema, report_query | Aggregated analytics | read-only |
datastore_list, datastore_query | Search Business Context documents | read-only |
integrations_list, integration_mcp_tool_call | Use connected apps (accounting, ERP, email) | read-only / write |
task_list, task_complete | Approvals and human tasks | read-only / write |
Every tool has a title and read-only, destructive and open-world hints, so your assistant can ask before it changes anything. The tools you see follow your workspace role and permissions.
Example prompts
- "Every Monday at 9am Singapore time, have someone in finance send me the AR ageing summary."
- "Hand off this vendor onboarding: collect the W-9 and bank details from Acme Ltd."
- "Which of my workflows failed last week?"
- "Show unpaid bills by vendor this quarter."
- "What does our expense policy say about client dinners?"
Privacy, security and disconnecting
- We store only hashes of tokens. Access tokens last 1 hour, and connections last 30 days.
- To revoke access at any time, go to app.peakflo.ai → Settings → MCP → Connected apps.
- Privacy policy · Terms · Disconnect and delete your data
Support
https://help.peakflo.co · [email protected]
Licence
The configuration files and docs in this repository are MIT licensed. The hosted service is covered by the Terms.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.