System 1 agents · House #22,239

What it is

Calls itself “System 1 decision models”.

• Calls itself: System 1 decision models

“System 1 decision models”

Where it puts itself

Its one line of difference: “Up to 6× faster and 25× cheaper than an LLM, at the same score.”

“Up to 6× faster and 25× cheaper than an LLM, at the same score.”

Who it sells to

Sells to developers.

• Sells to: developers

“…codex mcp add s1a -- uv run --project /path/to/system1-agents s1a-mcp”

How it runs

A API for Windows and macOS. Can be self-hosted. Offers an API. Open source. AI is the core of the product. You can bring your own API key.

• Kind: api
• Runs on: Windows · macOS
• Runs where: self-hosted
• API: yes
• Open source: yes
• AI role: core to the product
• Bring your own key: yes

“As an MCP server”

“Windows or macOS app window through Cua Driver”

What it works with

Names Claude Code, Codex and 3 more on its pages.

• Names: Claude Code · Codex · Cua-S1 Nano · +2

“Claude Code or Codex runs a prebuilt System 1 agent or builds a new one, on models like Jev, Laya or Cua-S1 Nano.”

“The caller skill, skills/s1a/SKILL.md , tells the host when to delegate and which command to run. It follows the agentskills.io layout. Codex also reads it through the .agents/skills/ symlink, and Hermes takes the same directory into ~/.hermes/skills .”

What it offers as proof

Live, no waitlist. Claims “6× faster and 25× cheaper”. Try it without signing up. Publishes docs.

• Number claimed: 6× faster and 25× cheaper
• Instant signup: yes
• Trial asks for: nothing

“Sign in”

“…git clone https://github.com/ThinkFlowLab/system1-agents && cd system1-agents”