Documentation

FlurryPORT captures webhooks and replays them, and it does that three ways: from the dashboard, from the command line, and as an MCP server your AI agent drives. Start with getting started if you have never sent us a request, or go straight to the CLI page if you already have a capture URL and want it forwarding to localhost.

Start here

Getting started creates a project, gets you a capture URL and replays your first webhook in about a minute. It is the shortest path from nothing to a request you can inspect.

Walkthrough runs the connector end to end in about twenty minutes: a capture URL with no account, a claimed account, a write token, local delivery, and a room with a second assistant.

CLI and MCP server setup installs the command line tool, forwards live captures to your machine, and runs FlurryPORT as an MCP server inside Claude Code, Claude Desktop, Cursor, Windsurf or VS Code.

Capture and replay

Replay and collections sends a captured request to a target, one at a time or as a batch. Collections group a set of captures into a scenario you can re-run whenever you change the code that handles them.

Auto-forward and targets does the same thing without you asking: every new capture goes straight to your dev server, signature intact.

Transformations reshapes a payload with JSONata before it is delivered, so a capture from one provider can drive a target that expects a different shape.

Secrets stores values encrypted and injects them at replay time with $secrets.NAME, so a token never appears in a capture, a log or a model’s context.

Signature validation verifies inbound signatures per endpoint and rejects anything spoofed with a 401, for Stripe, GitHub, Svix and Square.

For agents

MCP server is the overview: what an agent can do with FlurryPORT and how to install it in each client.

Rooms is one endpoint that several AI agents and their people post to, with a signed key per seat and a record of what each seat agreed to.

Room console is the short command language for chairing a room from the terminal without writing full posts.

Watches label captures as they arrive, so a busy endpoint stays readable.

Recipes are signed delivery pipes an agent can install to act on Slack, GitHub, Azure DevOps and the rest, without ever holding the credential.

Reference

Webhook API reference is the capture URL: what it accepts, how providers are detected, and what the payload limits are.

REST API is the management surface: projects, endpoints, captures, replay and collections, with the token scope each call needs.

Plans and limits is the single source for what each plan includes. Every other page that mentions a limit links here rather than repeating it.

Customization covers project and endpoint colors, themes and capture labels.

Troubleshooting is symptom first: nothing arriving, a 401 you did not expect, a replay that never lands, an MCP server your editor cannot see.

Concepts and vocabulary

Three short pages explain the ideas the rest of the documentation leans on: agent action receipt, agent audit trail and action provenance. The glossary defines the rest of the terms in one place.