MCP for AI agents

elcapture ships an MCP server. Point an AI coding agent at it and the agent can read the live request list, pull a full request, export a HAR, send a request and change rules — the same local API the CLI uses, over the Model Context Protocol.

Add it to Claude Code

claude mcp add elcapture -- elcapture-cli mcp

In Cursor, add an MCP server whose command is elcapture-cli mcp. Either way the app must be running, since the server talks to it on 127.0.0.1.

Tools the agent gets

Tool	Input	Does
`list_flows`	`filter`, `limit`	List captured requests.
`get_flow`	`id`	One request and response as a HAR entry (credentials redacted).
`export_har`	`filter`	Export captured requests as HAR.
`status`	—	Proxy address, request count, whether recording.
`clear`	—	Clear all captured requests.
`send_request`	`raw`	Send a raw HTTP request through the proxy.
`get_rules`	—	The current rules as JSON.
`set_rules`	`rules`	Replace all rules with a JSON array.

In practice

› what failed in the last run?      → list_flows
› show me that login request as JSON  → get_flow
› block analytics while I test        → set_rules

Captured bodies come from the network, so the server tells the agent to treat them as data, not instructions.