Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/Emanuele-web04/synara/llms.txt

Use this file to discover all available pages before exploring further.

Synara’s Cursor provider connects to the cursor-agent CLI — the headless agent runtime that Cursor exposes for programmatic use. Running it through Synara gives you the same Cursor agent capabilities within Synara’s local-first workspace: persistent threads, checkpoints, approval flows, and skill discovery.
This provider uses Cursor’s Agent Calling Protocol (ACP) interface. It is distinct from the Cursor desktop editor; you do not need the editor installed, only the cursor-agent CLI binary.

​
Prerequisites

The cursor-agent binary must be installed and either on your PATH or pointed to explicitly via binaryPath. No API key configuration is required inside Synara — the agent binary handles its own authentication with the Cursor backend.

​
Configuration

Provider settings live in the providers.cursor block of your Synara settings.json. See Configuration: Settings for the full settings file reference.
​
enabled
boolean
default:"true"
Enables or disables the Cursor provider. When false, Synara will not invoke the cursor-agent binary.
​
binaryPath
string
default:"cursor-agent"
Path to the cursor-agent executable. Accepts a bare binary name resolved from PATH or an absolute path. Change this if the binary lives outside your standard PATH.
​
apiEndpoint
string
default:""
Override the API endpoint the agent communicates with. Leave empty to use the default Cursor endpoint. Useful for custom enterprise deployments or local proxies.
​
customModels
string[]
Additional model slugs to offer in the model picker on top of the built-in Cursor model list.

​
Built-in models

Synara ships with the following Cursor model slugs pre-configured:
SlugName
autoAuto
composer-2Composer 2
claude-opus-4-6Claude Opus 4.6
gpt-5.3-codexGPT-5.3 Codex
gemini-3-proGemini 3 Pro
The default model when creating a new Cursor thread is auto.

​
Skills discovery

Cursor exposes reusable agent skills — domain-specific capabilities defined in Agent Skill folders. Synara discovers these automatically by scanning the same root directories that cursor-agent uses natively. The discovery logic (discoverCursorSkills) searches:
  • Project-local roots — .cursor/skills folders walking up from the current working directory.
  • User-global roots — ~/.cursor/skills-cursor and ~/.cursor/skills in the user’s home directory.
Discovered skills appear in Synara’s unified skill catalog and can be toggled per-thread or disabled globally from Configuration: Settings under skills.disabled.
Skills are identified by their lowercased name across providers. If the same skill name exists in both a Cursor folder and another provider’s catalog, Synara deduplicates them in the unified catalog.

​
Example settings.json

{
  "providers": {
    "cursor": {
      "enabled": true,
      "binaryPath": "cursor-agent"
    }
  }
}

​
Related

Build docs developers (and LLMs) love

Get started for freeTalk to us