commit

OpenClacky

Contributing? Read CONTRIBUTING.md before opening a PR.

The most Token-efficient open-source AI Agent.

OpenClacky matches Claude Code on capability at comparable cost, and saves significantly against other open-source agents (~50% vs OpenClaw, costs only ~1/3 of Hermes). 100% open source (MIT), BYOK with any OpenAI-compatible model, built on two years of Agentic R&D and harness engineering.

Website: https://www.openclacky.com/ · Backed by MiraclePlus · ZhenFund · Sequoia China · Hillhouse Capital

Why OpenClacky?

Same task, how much do you pay? Under comparable agent workloads, OpenClacky saves a large amount of Token spend compared to mainstream alternatives.

Numbers are averages measured on internal common agent tasks, using Claude Code as the baseline. Full benchmark reports will be published on GitHub.

Feature comparison

Core agent capability is roughly on par across the field — the real differentiators are cost, openness, Skill evolution, and integrations.

How we get the cost down

Not by cutting features — by compounding the right choice at every layer.

1. Ultra-high cache hit rate

Sessions never restart, double cache markers, Insert-then-Compress — the system prompt is never mutated, so compression still reuses the cache. Measured cache hit rate: near 100%.

2. Minimal tool set

Only 16 core tools. Capabilities are offloaded to the Skill ecosystem via a single invoke_skill meta-tool. Tool count is not the metric — task completion rate is.

3. Idle-time auto-compression

Go to a meeting, grab coffee — the agent compresses long context in the background and pre-warms the cache. Your first message back hits the cache directly. Cold-start first-token cost reduced by 50%+.

4. BYOK — you pick the model, you set the cost

Any OpenAI-compatible API, plug and play. Official direct, aggregate routing, compatible relays — the choice is 100% yours. Use Claude for code, auto-route subtasks to DeepSeek, save another chunk of tokens.

Built on 2 years · 3 generations of agentic architecture · 6 core harness engineering decisions.

Skills — the soul of the agent

• Invoke with / — instant browse, fuzzy search, direct call. Hundreds of Skills at your fingertips.
• Create Skills in natural language — just describe what you want; the agent drafts SKILL.md, breaks down steps, and runs validation. No code required.
• Self-evolving — after each run, the agent updates the Skill based on execution context and results. The next call is more stable and more accurate.
• Open & compatible — supports Claude Skills / Markdown Pack / custom formats.
• Monetizable — polished Skills can be packaged for sale, with encrypted distribution, License management, and creator-defined pricing.

Installation

Desktop installer (recommended)

Double-click to install — environment, dependencies, and Skills all set up automatically.

• macOS — Download .dmg (Apple Silicon / Intel)
• Windows — Download .exe (Windows 10 2004+ / Windows 11)

More options: https://www.openclacky.com/

Command line

One-line install(Mac/Ubuntu):

/bin/bash -c "$(curl -sSL https://raw.githubusercontent.com/clacky-ai/openclacky/main/scripts/install.sh)"

Windows:

powershell -c "& ([scriptblock]::Create((irm 'https://raw.githubusercontent.com/clacky-ai/openclacky/main/scripts/install.ps1')))"

or using Ruby(3.x/4.x):

Requirements: Ruby >= 3.1.0

gem install openclacky

see more: https://www.openclacky.com/docs/installation

Docker

Pre-built image (GHCR)

Images are published to GitHub Container Registry on version tags (v*).

# Replace <owner> with the repo owner (e.g. clacky-ai or your fork)
docker pull ghcr.io/<owner>/openclacky:latest
# or pin a release:
# docker pull ghcr.io/<owner>/openclacky:1.5.3

Linux:

docker run -d --name openclacky --network=host \
  -e CLACKY_ACCESS_KEY="" \
  -v openclacky-data:/root/.clacky \
  ghcr.io/<owner>/openclacky:latest

--network=host is required so the agent inside the container can reach Chrome's remote debugging port running on the host.

macOS / Windows:

docker run -d --name openclacky -p 7070:7070 \
  -e CLACKY_ACCESS_KEY="" \
  -v openclacky-data:/root/.clacky \
  ghcr.io/<owner>/openclacky:latest

Note: On macOS/Windows, --network=host is not supported — browser automation may be limited.

Open http://localhost:7070 after starting.

Build from source

git clone https://github.com/clacky-ai/openclacky.git
cd openclacky
# optional: set the OCI image version label
docker build --build-arg VERSION=1.5.3 -t openclacky .
docker run -d -p 7070:7070 -e CLACKY_ACCESS_KEY="" openclacky

Environment variables:

Quick Start

Terminal (CLI)

openclacky            # start interactive agent in current directory

Web UI

openclacky server     # default: http://localhost:7070

Open http://localhost:7070 for a full chat interface with multi-session support — run coding, copywriting, research sessions in parallel.

Options:

openclacky server --port 8080        # custom port
openclacky server --host 0.0.0.0     # listen on all interfaces (remote access)

Linux sandbox hosts can opt into task limits with openclacky server --task-cgroup /sys/fs/cgroup/tasks. The host must first create that cgroup, configure its aggregate memory/CPU limits, and enable the memory and cpu subtree controllers. Terminal shells and browser MCP daemons enter child groups before exec; the server remains outside. Terminal results include resource_exhausted: "memory" and an error when their group records an OOM kill, even if the shell masks the child's failure. A separately managed Chrome must also be placed under the same parent by its launcher. This option is off by default and does not establish a security boundary against privileged commands.

Configuration

$ openclacky
> /config
``