AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
翻译准备中
尚未检测到更新。GitHub 发布版本后会显示在这里。
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证Eaon Code — Eaon's official coding agent. Token-efficient, extensible, built atop the Pi harness.
翻译准备中
取自仓库 README
正在加载完整 README。
查看完整 README ↗Eaon Code is a full-screen coding agent for the terminal, with multi-provider model access, durable sessions, and an extension system for adapting it to a project or team.
/plan) restricts the agent to read-only exploration and shows a small mode indicator./swarm) makes the parent orchestrate 2–6 named specialist sub-agents in parallel, then reconcile and verify their work./agents to inspect conversations, steer work, or press x twice to stop a run; completed results expand with Ctrl+O.~/.eaon/agent and project .eaon/ directories. Existing .pi configuration directories remain usable as a compatibility path; Eaon Code does not move or delete them.The core agent also includes file and shell tools, streaming responses, session branching and compaction, provider login and API-key auth, skills, prompts, extensions, themes, package management, and print/JSON/RPC/SDK interfaces.
Eaon Code includes BetterWright's official Pi browser integration as a built-in feature. The browser tools are ready in the agent without installing BetterWright as a separate extension. BetterWright keeps a browser session across calls and applies its network and browser policies while Eaon navigates and interacts with pages.
On first use, install BetterChromium once with Bun 1.4 or newer:
bunx betterwright setup
See the BetterWright setup guide and package page for the integration details.
One line:
curl -fsSL https://raw.githubusercontent.com/eaonlabs/eaon-code/main/install.sh | bash
The installer clones the repository into ~/.local/share/eaon-code and builds it with Bun. It needs Git and Bun on PATH.
The CLI command is eaon-code:
export ANTHROPIC_API_KEY=sk-ant-... # or another provider
eaon-code
Or launch and log in interactively:
eaon-code
/login
Local run from a private clone (no publish required):
git clone https://github.com/eaonlabs/eaon-code.git
cd eaon-code
bun install --frozen-lockfile --ignore-scripts
bun run try:local -- --help # Build the checkout and run its CLI without installing it
| Package | Description |
|---|---|
| @eaonlabs/eaon-code | Interactive coding agent CLI |
| @eaonlabs/eaon-agent-core | Agent runtime with tool calling and state management |
| @eaonlabs/eaon-ai | Unified multi-provider LLM API (OpenAI, Anthropic, Google, …) |
| @eaonlabs/eaon-tui | Terminal UI library with differential rendering |
| @eaonlabs/eaon-telemetry | Telemetry contracts, adapters, and typed schemas |
| @eaonlabs/chord | Application-composition runtime for services, RPC, and plugins |
New installs use ~/.eaon/agent/ for user settings, credentials, sessions, and resources, and .eaon/ for project-local settings and resources. Existing .pi directories are still recognized when no corresponding .eaon directory exists, so upgrading does not strand or rewrite existing data.
Eaon Code does not include a built-in permission system for restricting filesystem, process, network, or credential access. By default it runs with the permissions of the user and process that launched it.
If you need stronger boundaries, containerize or sandbox it. See packages/coding-agent/docs/containerization.md for three patterns:
! commands into a local Linux micro-VM.bun install --frozen-lockfile --ignore-scripts # Install all dependencies without running lifecycle scripts
bun run build # Refresh model data, then build all packages
bun run try:local -- --help # Build and run this checkout without installing it
bun run build:offline # Rebuild using existing model data without network access
bun run check # Lint, format, and type check
./test.sh # Run tests (skips LLM-dependent tests without API keys)
See CONTRIBUTING.md and AGENTS.md.
MIT