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 drives the OpenAI Codex CLI as a local child process, giving you a full AI coding session inside the Synara workspace without any intermediary server. Codex is the default provider in Synara — when you create a new thread, it uses Codex unless you pick a different provider. Because Synara communicates with the Codex CLI over its JSON-RPC interface, the CLI must be installed on your machine and authorized with your OpenAI account before Synara can use it.

​
Prerequisites

  • Codex CLI installed — see the official repository for installation instructions.
  • Codex CLI authorized with a valid OpenAI API key or OAuth session (run codex once in your terminal to complete the login flow).
  • Minimum supported version: 0.37.0. Synara checks the installed version on startup and will warn you if it is too old.
Synara cannot install or update the Codex CLI for you. If the binary is missing or below v0.37.0, Synara will display an upgrade prompt in the provider status panel. Run npm install -g @openai/codex (or your preferred install method) to upgrade, then restart Synara.

​
Configuration

Codex is configured under providers.codex in settings.json. All fields are optional — Synara applies sensible defaults when a field is absent. 
{
  "providers": {
    "codex": {
      "enabled": true,
      "binaryPath": "codex",
      "homePath": "",
      "customModels": []
    }
  }
}
​
enabled
boolean
default:"true"
Set to false to disable the Codex provider entirely. Synara will skip Codex during startup discovery and will not allow new Codex threads.
​
binaryPath
string
default:"codex"
The command or full path used to launch the Codex CLI. Defaults to "codex", which resolves via your PATH. Use an absolute path (e.g. /home/user/.local/bin/codex) if the binary is not on your PATH or if you need to pin to a specific installation.
​
homePath
string
default:""
An optional override for the home directory Codex uses when it starts. When left empty Codex uses its default home resolution. Set this if you have multiple Codex configurations and need Synara to use a specific one — for example, a project-specific credentials directory.
​
customModels
string[]
default:"[]"
A list of additional model slugs to show in the model picker alongside Synara’s built-in Codex model list. Each entry is a plain string (e.g. "gpt-5.5-preview"). Useful for early-access or custom model endpoints.
Set homePath if you keep multiple Codex authentication profiles on the same machine and need Synara to use a non-default home directory — for instance, a directory with credentials scoped to a specific organization.

​
Model options

When you start or configure a Codex thread you can tune how the model reasons. These options are passed directly to the Codex CLI at session start. 
// From packages/contracts/src/model.ts
export const CODEX_REASONING_EFFORT_OPTIONS = ["low", "medium", "high", "xhigh"] as const;

export const CodexModelOptions = Schema.Struct({
  reasoningEffort: Schema.optional(TrimmedNonEmptyString),
  fastMode: Schema.optional(Schema.Boolean),
});
​
reasoningEffort
string
Controls how much reasoning the model applies before replying. Built-in values are "low", "medium", "high", and "xhigh". Codex runtime discovery can expose early-access effort values outside this set; any non-empty string is accepted.Defaults per model:
  • GPT-5.5 — "medium"
  • GPT-5.4, GPT-5.4 Mini, GPT-5.3 Codex, GPT-5.3 Codex Spark, GPT-5.2 Codex, GPT-5.2 — "high"
​
fastMode
boolean
When true, instructs the Codex CLI to prefer lower-latency responses. Availability depends on the selected model.

​
Built-in model roster

The following Codex models ship with Synara’s built-in model list. Additional models can be added via customModels.
SlugName
gpt-5.5GPT-5.5
gpt-5.4GPT-5.4
gpt-5.4-miniGPT-5.4 Mini
gpt-5.3-codexGPT-5.3 Codex
gpt-5.3-codex-sparkGPT-5.3 Codex Spark
gpt-5.2-codexGPT-5.2 Codex
gpt-5.2GPT-5.2
The default model for new Codex threads is gpt-5.5.

​
Related

Build docs developers (and LLMs) love

Get started for freeTalk to us