Quorum

Quorum is a Rust-first coding agent CLI with first-class multi-provider support.

It supports OpenAI/Codex, Anthropic, Grok/xAI, Gemini, Azure OpenAI, and more, while still preserving provider-specific strengths instead of flattening everything into a lowest-common-denominator abstraction.

Public distribution:

• Releases: https://github.com/xudesheng/quorum/releases
• Latest release: https://github.com/xudesheng/quorum/releases/latest
• Install guide: install.md
• Config reference: config.md

Quick Install

Fastest path:

npm i -g @xudesheng/quorum
quorum --version

For all install methods (npm, GitHub binaries, Homebrew, winget, bunx), see install.md.

Quick Start

Recommended config path:

• ~/.quorum/config.toml

Full config reference: config.md

Example config (OpenAI + Anthropic + Grok + Gemini + Azure OpenAI + Codex OAuth):

[defaults]
config_item = "openai-prod"

[[config_items]]
id = "openai-prod"
provider = "openai_responses"
description = "OpenAI with inline secret example"

[config_items.auth]
mode = "api_key"
secret = "sk-proj-REPLACE_ME"

[config_items.model_policy]
default_model = "gpt-5.3-codex"
allowed_models = ["gpt-5.3-codex"]
mutable = false

[[config_items]]
id = "anthropic-prod"
provider = "anthropic"

[config_items.auth]
mode = "api_key"
env_var = "ANTHROPIC_API_KEY"

[config_items.model_policy]
default_model = "claude-opus-4-6"
allowed_models = ["claude-opus-4-6"]
mutable = false

[[config_items]]
id = "grok-prod"
provider = "grok"

[config_items.auth]
mode = "api_key"
env_var = "XAI_API_KEY"

[config_items.model_policy]
default_model = "grok-4-1-fast-reasoning"
allowed_models = ["grok-4-1-fast-reasoning"]
mutable = false

[[config_items]]
id = "gemini-prod"
provider = "gemini"

[config_items.auth]
mode = "api_key"
env_var = "GEMINI_API_KEY"

[config_items.model_policy]
default_model = "gemini-3-flash-preview"
allowed_models = ["gemini-3-flash-preview"]
mutable = false

[[config_items]]
id = "azure-openai-prod"
provider = "azure_openai"

[config_items.auth]
mode = "api_key"
env_var = "AZURE_OPENAI_API_KEY"

[config_items.endpoint]
type = "custom"
base_url = "https://YOUR_RESOURCE.openai.azure.com/openai/v1"
deployment = "gpt-5-3-codex-deploy"
api_version = "2025-04-01-preview"

[config_items.model_policy]
default_model = "gpt-5.3-codex"
allowed_models = ["gpt-5.3-codex"]
mutable = false

[[config_items]]
id = "codex-oauth-dev"
provider = "openai_responses"

[config_items.auth]
mode = "oauth"

[config_items.model_policy]
default_model = "gpt-5.3-codex"
allowed_models = ["gpt-5.3-codex"]
mutable = false

Notes:

• Anthropic OAuth is currently restricted by Anthropic. Use api_key for Anthropic.
• Gemini uses API key auth only in Quorum; configure GEMINI_API_KEY.
• For API key auth, configure either secret or env_var; only one is required.
• defaults.config_item is the convenience entry point for daily use.

Supported Features

1) MCP Server

Quorum supports MCP servers over stdio and streamable-http.

Example (.quorum/mcp.toml):

[mcp-servers.mcp_test]
transport = "stdio"
command = "/Users/you/.local/bin/uv"
args = ["run", "/path/to/minimal_server.py"]
enabled = true
allow_indirect_launcher = true

2) Skills

Quorum supports built-in starter skills and custom SKILL.md skills.

Skill locations:

• User scope: ~/.quorum/skills/<skill-name>/SKILL.md
• Project scope: <workspace>/.quorum/skills/<skill-name>/SKILL.md

Minimal SKILL.md:

---
name: commit_helper
version: 1.0.0
trigger:
  slash: /commit-helper
---

## setup
You help draft concise conventional commit messages.

3) Auto-Learn

Quick docs:

• English: auto-learn.md
• 中文: auto-learn_CN.md

Samples:

• samples/AUTO-LEARNED.md (sample index)
• samples/learned/debugging.md (sample topic)

Common setup:

• ~/.quorum/AUTO-LEARNED.md
• ~/.quorum/learned/*.md

4) Steering Message

Quorum supports runtime steering/follow-up while a turn is running:

• Enter: steer
• Tab: follow-up
• Esc: interrupt

This is useful for correcting direction during long responses or tool execution.