Skip to content

Commit af9d9e7

Browse files
committed
Prepare repository for public beta
Change-Id: Ic22af3a5fd2d05adf772f793d4872eeb6e1d4b65
1 parent 3ab4092 commit af9d9e7

11 files changed

Lines changed: 161 additions & 1415 deletions

File tree

CHANGELOG.md

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,21 @@
1+
# Changelog
2+
3+
All notable project-level changes are documented here. Package-specific details remain in each
4+
package's own changelog.
5+
6+
## [Unreleased]
7+
8+
- Prepare the repository, documentation, and community surfaces for the public beta.
9+
10+
## 0.0.2-beta.0 — 2026-07-13
11+
12+
First coordinated beta release of:
13+
14+
- `@openagentpack/cli`
15+
- `@openagentpack/sdk`
16+
- `@openagentpack/playground`
17+
18+
The beta introduces the declarative `agents.yaml` workflow, provider capability contracts,
19+
planning and drift detection, managed resource lifecycle commands, and the local Playground.
20+
21+
Beta releases may contain breaking API and configuration changes before `1.0`.

NOTICE

Lines changed: 30 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -51,3 +51,33 @@ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
5151
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
5252
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
5353
THE SOFTWARE.
54+
55+
--------------------------------------------------------------------------------
56+
mattpocock/skills — https://github.com/mattpocock/skills
57+
--------------------------------------------------------------------------------
58+
59+
The repository includes development-agent skills derived from this project
60+
under `.agents/skills/`, with source paths and content hashes recorded in
61+
`skills-lock.json`.
62+
63+
MIT License
64+
65+
Copyright (c) 2026 Matt Pocock
66+
67+
Permission is hereby granted, free of charge, to any person obtaining a copy
68+
of this software and associated documentation files (the "Software"), to deal
69+
in the Software without restriction, including without limitation the rights
70+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
71+
copies of the Software, and to permit persons to whom the Software is
72+
furnished to do so, subject to the following conditions:
73+
74+
The above copyright notice and this permission notice shall be included in all
75+
copies or substantial portions of the Software.
76+
77+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
78+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
79+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
80+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
81+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
82+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
83+
SOFTWARE.

README.md

Lines changed: 25 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,12 @@
88
99
[![License](https://img.shields.io/badge/license-Apache_2.0-blue.svg)](./LICENSE)
1010
[![CI](https://github.com/modelstudioai/OpenAgentPack/actions/workflows/ci.yml/badge.svg)](https://github.com/modelstudioai/OpenAgentPack/actions/workflows/ci.yml)
11+
[![npm](https://img.shields.io/npm/v/@openagentpack/cli?label=npm&color=cb3837)](https://www.npmjs.com/package/@openagentpack/cli)
1112
[![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](./CONTRIBUTING.md)
1213

14+
> [!IMPORTANT]
15+
> OpenAgentPack is in beta. Public APIs and the `agents.yaml` schema may change before `1.0`. See the [changelog](./CHANGELOG.md).
16+
1317
<p align="center">
1418
<img src="https://img.alicdn.com/imgextra/i3/O1CN01xWDp5P1EP90HOZx9q_!!6000000000343-2-tps-1254-1254.png" width="360" alt="OpenAgentPack: one agents.yaml for multiple managed-agent providers">
1519
</p>
@@ -22,40 +26,44 @@ One `agents.yaml` defines an agent's environment, model, instructions, tools, sk
2226

2327
```bash
2428
npm install -g @openagentpack/cli
25-
agents init && agents validate && agents plan
29+
agents init
30+
# Configure one provider's credentials, then:
31+
agents validate && agents plan
2632
```
2733

28-
[Run the 5-minute quick start](./docs/getting-started.md) · [View provider support](./docs/reference/providers.md) · [Browse runnable examples](./docs/examples.md)
34+
[Run the 5-minute quick start](./docs/getting-started.md) · [View provider support](./docs/reference/providers.md) · [Browse runnable examples](./docs/examples.md) · [Roadmap](./ROADMAP.md)
2935

3036
## Why now
3137

32-
Agent harnesses are converging on a stable shape — environment, vault, memory, skills, files, MCP servers, prompt, agent loop, multi-agent orchestration. Two shifts ride alongside: agents moving from local runtimes to remote managed ones, and AI-native teams moving from individual productivity to organizational productivity. What's still missing is **reliable, portable agent infrastructure** that doesn't lock you to one provider.
38+
Agents are moving from personal tools to enterprise digital workers. But the things that make an agent valuable — its prompts, skills, knowledge files, tools, and runtime configuration — still live mainly inside cloud-provider consoles.
39+
40+
These are business assets. They should be managed, reviewed, handed over, reproduced, and migrated like code, data, and documents — not trapped as a pile of clicks in one console.
3341

3442
## Why OpenAgentPack
3543

36-
You built an agent somewhere — a prompt, tools, skills, knowledge files, credentials, runtime settings. Those are your business assets, but today they live as a pile of clicks inside one vendor's console: impossible to review in a PR, roll back, reproduce, or move. As agents move from "call a model API in your own code" to "assemble an agent inside a vendor console", vendor lock-in re-forms **one layer up** — at the managed harness. OpenAgentPack is a bet against that.
44+
OpenAgentPack puts a declarative control plane between the agent and the cloud platform. The enterprise owns the declaration; provider adapters render it into real managed agents on Bailian, Qoder, Claude, or Volcengine Ark.
3745

38-
OpenAgentPack turns those assets into a portable agent definition that deploys to different agent platforms. Three things this gives you:
46+
The goal is to make an agent an enterprise-controlled, portable, and inheritable digital asset.
3947

40-
### No platform lock-in
48+
### Declaration and portability: the agent as a blueprint
4149

42-
Your agent isn't locked to one platform. Today on Bailian, tomorrow on Qoder / Claude / Volcengine Ark — and more to come.
50+
Borrowing Docker's declarative idea, OpenAgentPack brings everything that determines what an agent is — model, instructions, tools, skills, environment, files, and credential references — into one `agents.yaml` blueprint. The blueprint can live in Git, pass through pull-request review, reproduce an agent, and move across providers.
4351

44-
### Reproduce, migrate, compare
52+
### State and governance: plan before execution
4553

46-
The same agent can run on multiple platforms; see which is cheaper, better, or faster.
54+
Borrowing Terraform's state-driven workflow, OpenAgentPack keeps desired config, local state, and remote state distinct. `plan` previews creates, updates, and deletes; `apply` executes them in dependency order; drift detection finds console-side changes; and a previous declaration can restore a known-good configuration.
4755

48-
### Your agent is your asset
56+
### Validation and experience: Playground as the showroom
4957

50-
Prompts, tools, skills, knowledge files, configuration — no longer just a pile of clicks in a console, but something you can save, reuse, review, and hand off.
58+
Even a precise blueprint needs to be experienced. Playground runs real sessions from the same declaration and lets teams exercise the same scenario against different providers. Provider comparison becomes an observable result, not only a capability matrix.
5159

52-
Focus on business innovation, not on babysitting agent infra. Switch providers the way Docker lets you switch hosts, and keep your definition intact across the move.
60+
> OpenAgentPack uses a Docker-like declaration to draw the agent blueprint, a Terraform-like state model to manage construction and acceptance, and Playground as the showroom — so enterprises can manage agents the way they manage code.
5361
54-
The rest is mechanics: a single `agents.yaml`, a `validate → plan → apply` workflow, content-hash diffing, dependency-aware ordering, drift recovery. The YAML is always the single source of truth. See [Agents as code](./docs/concepts/agents-as-code.md) for the mental model and [CONTEXT.md](./CONTEXT.md) for the precise vocabulary (agent harness vs. agent infra, capability contract).
62+
The mechanics are a single `agents.yaml`, a `validate → plan → apply` workflow, content-hash diffing, dependency-aware ordering, and drift recovery. The YAML remains the source of truth. See [Agents as code](./docs/concepts/agents-as-code.md) for the mental model and [CONTEXT.md](./CONTEXT.md) for the precise vocabulary.
5563

5664
- **Declarative** — one `agents.yaml` describes your whole agent stack. Commit it, review it in a PR, roll it back.
5765
- **Terraform-style workflow**`validate → plan → apply`. Preview every create / update / delete before it happens.
58-
- **Multi-provider** — the same agent definition deploys to Bailian, Qoder, Claude, or Volcengine Ark. Switch vendors by changing two lines.
66+
- **Multi-provider**reuse the core declaration across Bailian, Qoder, Claude, and Volcengine Ark; the [capability contract](./docs/reference/providers.md) makes native, emulated, and unsupported differences explicit.
5967
- **Incremental** — content-hash diffing updates only what actually changed; no redundant API calls.
6068
- **Dependency-aware** — Environment → Skill → Agent are created in topological order; a failed dependency skips its dependents instead of leaving half-built state.
6169
- **Drift recovery** — detects when remote config has drifted from your declaration and reconciles it. The YAML is always the single source of truth.
@@ -70,7 +78,7 @@ The rest is mechanics: a single `agents.yaml`, a `validate → plan → apply` w
7078

7179
`agents playground` launches a local WebUI, fetches the matching `@openagentpack/playground` package on demand, and opens it in your browser. Use `--provider` to target `bailian`, `qoder`, `ark`, or `claude`.
7280

73-
[Watch the Playground demo video](https://cloud.video.taobao.com/vod/f9cVQvN8vYeW2YfRZ59qv5SgJUDgsm-r48mpKIB0Has.mp4)
81+
[Watch the Playground demo video](https://github.com/user-attachments/assets/bf51b8d8-f2ed-464b-bca9-0709fefcc44d)
7482

7583
## Quick start
7684

@@ -192,6 +200,8 @@ Or launch a packaged local UI with `agents playground --provider <bailian|qoder|
192200

193201
Contributions are welcome. Please read [CONTRIBUTING.md](./CONTRIBUTING.md) for the dev setup, merge requirements, and how to add a new provider. All participants are expected to follow our [Code of Conduct](./CODE_OF_CONDUCT.md).
194202

203+
Use [GitHub Discussions](https://github.com/modelstudioai/OpenAgentPack/discussions) for questions and design proposals, and [GitHub Issues](https://github.com/modelstudioai/OpenAgentPack/issues) for reproducible bugs and accepted work. Current priorities are tracked in the [public roadmap](./ROADMAP.md).
204+
195205
## Security
196206

197207
Found a vulnerability? Please follow the process in [SECURITY.md](./SECURITY.md) — do not open a public issue.

README.zh-CN.md

Lines changed: 27 additions & 18 deletions
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,12 @@
88
99
[![License](https://img.shields.io/badge/license-Apache_2.0-blue.svg)](./LICENSE)
1010
[![CI](https://github.com/modelstudioai/OpenAgentPack/actions/workflows/ci.yml/badge.svg)](https://github.com/modelstudioai/OpenAgentPack/actions/workflows/ci.yml)
11+
[![npm](https://img.shields.io/npm/v/@openagentpack/cli?label=npm&color=cb3837)](https://www.npmjs.com/package/@openagentpack/cli)
1112
[![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](./CONTRIBUTING.md)
1213

14+
> [!IMPORTANT]
15+
> OpenAgentPack 目前处于 Beta 阶段。公开 API 和 `agents.yaml` Schema 在 `1.0` 前仍可能发生不兼容变更,详见 [更新日志](./CHANGELOG.md)
16+
1317
<p align="center">
1418
<img src="https://img.alicdn.com/imgextra/i1/O1CN016pw0Ax1KuBXckX0Iz_!!6000000001223-2-tps-1254-1254.png" width="360" alt="OpenAgentPack:一份 agents.yaml 管理多个托管 Agent 平台">
1519
</p>
@@ -22,40 +26,44 @@
2226

2327
```bash
2428
npm install -g @openagentpack/cli
25-
agents init && agents validate && agents plan
29+
agents init
30+
# 配置一个 Provider 的凭证后:
31+
agents validate && agents plan
2632
```
2733

28-
[5 分钟快速开始](./docs/getting-started.md) · [查看 Provider 支持](./docs/reference/providers.zh-CN.md) · [浏览可运行示例](./docs/examples.md)
34+
[5 分钟快速开始](./docs/getting-started.zh-CN.md) · [查看 Provider 支持](./docs/reference/providers.zh-CN.md) · [浏览可运行示例](./docs/examples.md) · [Roadmap](./ROADMAP.md)
2935

3036
## 为什么是现在
3137

32-
Agent Harness 形态正趋于稳定 —— environment、vault、memory、skills、files、MCP servers、prompt、agent loop、multi-agent orchestration。两个趋势并行:Agent 从本地走向远程托管,团队从个人效率走向组织效率(AI Native 组织)。缺的还是一个**稳定可靠、不被厂商锁定的 Agent Infra**
38+
Agent 正在从个人工具变成企业的数字员工。但企业真正重要的东西 —— Prompt、技能、知识文件、工具和运行配置 —— 现在主要只存在于云厂商控制台里。
39+
40+
这些是企业的业务资产。它们不该只活在某个控制台里,而应该像代码、数据和文档一样,可以被管理、被审查、被交接、被复现,也可以自由迁移。
3341

3442
## 为什么用 OpenAgentPack
3543

36-
你在某个平台上搭了一个 Agent —— 里面有 Prompt、工具、技能、知识文件、凭据、运行设置。这些东西其实是你的业务资产,但今天它们只是一堆控制台里的点击 —— 无法在 PR 里审查、无法回滚、无法复现、无法迁移。当 Agent 从"在自己代码里调模型 API"变成"在厂商控制台里组装 Agent",厂商锁定往**上一层**重新形成 —— 在 managed harness 层。OpenAgentPack 就是对这个趋势的对抗
44+
OpenAgentPack 在 Agent 和云平台之间建立一个声明式控制平面。企业拥有自己的声明,不同平台通过各自的 Adapter,把同一份声明渲染成百炼、Qoder、Claude 或火山方舟中的真实 Agent。
3745

38-
OpenAgentPack 把这些资产变成一份可携带的 Agent 定义,部署到不同的 Agent 平台。你得到三件事:
46+
我们的目标,是让 Agent 成为企业可掌控、可迁移、可传承的数字资产。
3947

40-
### 我的 Agent 不被某个平台锁死
48+
### 声明与可携带:把 Agent 写成一份图纸
4149

42-
今天在百炼,明天可以去 Qoder、Claude、火山方舟,未来也可以接更多平台
50+
借鉴 Docker 的声明思想,把所有决定 Agent 是什么的东西 —— 模型、指令、工具、技能、环境、文件和凭据引用 —— 收敛到一份 `agents.yaml` 图纸里。这份图纸可以进入 Git、接受 PR 审查、复现 Agent,也可以迁移到不同 Provider
4351

44-
### 我的 Agent 可以复制、迁移、对比
52+
### 状态与可治理:先计划,再执行
4553

46-
同一个 Agent 可以在多个平台跑,看哪个成本更低、效果更好、速度更快
54+
借鉴 Terraform 的状态驱动思想,明确区分期望配置、本地状态和远端状态。`plan` 告诉你即将创建、更新和删除什么;`apply` 按依赖拓扑顺序安全执行;控制台被手动修改后可以发现 Drift;回到之前的稳定声明即可重新收敛
4755

48-
### 我的 Agent 是我自己的资产
56+
### 验证与可体验:Agent 的样板间
4957

50-
Prompt、工具、技能、知识文件、配置,不再只是某个控制台里的一堆点击,而是可以保存、复用、审查、交接的东西
58+
图纸再精确,也不如走进样板间体验。Playground 从同一份声明运行真实 Session,让团队在不同 Provider 上执行同一个场景。多平台对比不再只是一张能力矩阵,而是可观察、可体验的真实结果
5159

52-
专注业务创新,无须单独人力维护 Agent Infra 基础设施。像 Docker 一样在不同平台间迁移,定义保持不变
60+
> OpenAgentPack 借鉴 Docker 的声明思想绘制 Agent 图纸,借鉴 Terraform 的状态驱动思想管理施工与验收,再用 Playground 作为样板间完成效果验证 —— 让企业像管代码一样管 Agent
5361
54-
剩下的都是机制:一个 `agents.yaml``validate → plan → apply` 工作流、内容哈希增量检测、依赖拓扑排序、Drift 恢复。YAML 始终是唯一事实来源。心智模型见 [Agents as code](./docs/concepts/agents-as-code.md),精确词汇见 [CONTEXT.md](./CONTEXT.md)(agent harness vs. agent infra、capability contract)
62+
具体机制包括一份 `agents.yaml``validate → plan → apply` 工作流、内容哈希增量检测、依赖拓扑排序和 Drift 恢复。YAML 始终是事实来源。心智模型见 [Agents as code](./docs/concepts/agents-as-code.md),精确词汇见 [CONTEXT.md](./CONTEXT.md)
5563

5664
- **声明式** —— 一个 `agents.yaml` 描述整套 Agent 基础设施。提交它、在 PR 里审查它、随时回滚它。
5765
- **Terraform 式工作流** —— `validate → plan → apply`,每一次 create / update / delete 执行前都能预览。
58-
- **多 Provider** —— 同一份 Agent 定义可部署到百炼、Qoder、Claude和火山方舟,换厂商只改两行
66+
- **多 Provider** —— 核心声明可复用于百炼、Qoder、Claude 和火山方舟;[能力契约](./docs/reference/providers.zh-CN.md) 会明确表达 native、emulated 和 unsupported 差异
5967
- **增量变更** —— 基于内容哈希检测差异,只更新真正变化的资源,不做无意义的 API 调用。
6068
- **依赖自动解析** —— Environment → Skill → Agent 按拓扑序创建;依赖失败会跳过下游,而不是留下半成品状态。
6169
- **Drift 恢复** —— 检测远程配置与声明的漂移并收敛回来。YAML 始终是唯一事实来源。
@@ -70,8 +78,7 @@ Prompt、工具、技能、知识文件、配置,不再只是某个控制台
7078

7179
`agents playground` 会启动本地 WebUI,按需拉取匹配版本的 `@openagentpack/playground`,并自动打开浏览器。可通过 `--provider` 指定 `bailian``qoder``ark``claude`
7280

73-
https://github.com/user-attachments/assets/bf51b8d8-f2ed-464b-bca9-0709fefcc44d
74-
81+
[观看 Playground 演示视频](https://github.com/user-attachments/assets/bf51b8d8-f2ed-464b-bca9-0709fefcc44d)
7582

7683
## 快速开始
7784

@@ -114,7 +121,7 @@ agents:
114121
builtin: [bash, read, glob, grep]
115122
```
116123
117-
密钥通过 `${VAR_NAME}` 语法引用,从 `.env` 加载 —— 绝不写进配置文件本身。完整流程见 [Getting started](./docs/getting-started.md)。
124+
密钥通过 `${VAR_NAME}` 语法引用,从 `.env` 加载 —— 绝不写进配置文件本身。完整流程见 [快速开始](./docs/getting-started.zh-CN.md)。
118125

119126
## 安装
120127

@@ -150,7 +157,7 @@ npm install -g @openagentpack/cli
150157

151158
| 文档 | 内容 |
152159
|------|------|
153-
| [Getting started](./docs/getting-started.md) | 从安装到跑通一个 session 的最短路径。 |
160+
| [快速开始](./docs/getting-started.zh-CN.md) | 从安装到跑通一个 session 的最短路径。 |
154161
| [配置指南](./docs/guides/configure-an-agent.zh-CN.md) | 从最简到完整的渐进式教程。 |
155162
| [配置参考](./docs/reference/configuration.md) | `agents.yaml` 全部字段、类型与说明。 |
156163
| [CLI 参考](./docs/reference/cli.md) | 所有 `agents` 命令、选项与行为。 |
@@ -193,6 +200,8 @@ bun run dev # 同时启动 server + webui
193200

194201
欢迎贡献。请阅读 [CONTRIBUTING.md](./CONTRIBUTING.md) 了解开发环境搭建、合并要求以及如何新增 Provider。所有参与者需遵守我们的 [行为准则](./CODE_OF_CONDUCT.md)。
195202

203+
使用 [GitHub Discussions](https://github.com/modelstudioai/OpenAgentPack/discussions) 提问或提交设计建议,使用 [GitHub Issues](https://github.com/modelstudioai/OpenAgentPack/issues) 报告可复现问题和认领已接受任务。当前优先级见 [公开 Roadmap](./ROADMAP.md)。
204+
196205
## 安全
197206

198207
发现漏洞?请按 [SECURITY.md](./SECURITY.md) 的流程处理 —— 不要公开提 issue。

0 commit comments

Comments
 (0)