cursor

A setup guide for connecting CEL, an agent execution system, to Cursor, a code editor with an AI chat and composer. CEL exposes actions through MCP, a standard way for tools to connect to AI applications.

In plain words
What is it for?
Use it to build CEL, add it to Cursor's MCP settings, and let Cursor call CEL actions from the editor.
Why use it?
It explains how Cursor handles planning and approvals while CEL handles execution, avoiding the need for custom integration code.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/dimpagk92/cellar/cursor
Clone the repo
git clone --depth 1 https://github.com/dimpagk92/cellar
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,365 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.01365
Opus 5 $0.00000 $0.00682
Sonnet 5 $0.00000 $0.00273
Haiku 4.5 $0.00000 $0.00136

Measured yesterday against content hash 75e4107aea58, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cursor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

docs/agents/cursor.md · 135 lines

How it starts

The opening of the file, as written. The whole thing — 135 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Driving CEL from Cursor

This page shows how to drive CEL from Cursor.

Read docs/adapters-cel-agents.md first if you haven't.

Purpose

Cursor supports MCP servers from its composer / chat runtime, which makes CEL usable directly from the editor without any bespoke glue. In this setup:

  • Cursor owns planning, tool selection, retries, and approval prompts.
  • CEL exposes cel_see, cel_act, cel_perceive, cel_think over stdio MCP.
  • You normally only need cel_see + cel_act from Cursor. Leave cel_think alone unless you explicitly want delegated autonomy.

Setup

1. Build CEL

cd /path/to/cellar
pnpm install && pnpm -r build

2. Register CEL as an MCP server

Cursor reads MCP servers from its global settings and/or project settings. The most reliable location is the global MCP config:

  • macOS: ~/.cursor/mcp.json

TODO: verify with latest Cursor version — Cursor has iterated on MCP config paths. If ~/.cursor/mcp.json does not exist for you, check Cursor → Settings → MCP (the UI now exposes a server editor in recent builds). The stdio config block itself is stable across versions.

Add CEL:

{
  "mcpServers": {
    "cel": {
      "command": "node",
      "args": ["/absolute/path/to/cellar/mcp-server/dist/index.js"]
    }
  }
}

Or, if cellar is on your PATH:

{
  "mcpServers": {
    "cel": {
      "command": "cellar",
      "args": ["mcp"]
    }
  }
}

3. Confirm tools appear

Restart Cursor. Open Settings → MCP (or the equivalent panel). You should see cel listed with its four tools: cel_see, cel_act, cel_perceive, cel_think.

TODO: screenshot — Cursor Settings panel with CEL tools listed.

In the composer, enabling the CEL tools in the tool picker makes them callable from the current session.

4. Grant macOS Accessibility permission

The first call to cel_act that performs input will fail until the host process (Cursor, or the Node process Cursor spawns) has Accessibility permission. Grant it in System Settings → Privacy & Security → Accessibility and restart Cursor.

Read the full file on GitHub · 135 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 135 lines · 0 tokens per session scan A 75e4107aea58

Subscribe to this mod's changes

cursor is an agent published in the GitHub repository dimpagk92/cellar (4 stars, last pushed 22d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 1,365 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.