You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+13-15Lines changed: 13 additions & 15 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -14,9 +14,9 @@
14
14
> [!IMPORTANT]
15
15
> OpenAgentPack is in beta. Public APIs and the `agents.yaml` schema may change before `1.0`. See the [changelog](./CHANGELOG.md).
16
16
17
-
<palign="center">
18
-
<imgsrc="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">
19
-
</p>
17
+

18
+
19
+
`agents.yaml → validate → plan → apply`: bring agents back into Git, review the pending changes, then update remote resources.
20
20
21
21
One `agents.yaml` defines an agent's environment, model, instructions, tools, skills, MCP servers, vaults, and credentials. Review every change in a PR, preview it with `plan`, and apply it when ready — instead of rebuilding the same agent through console clicks.
▶ [Watch the Playground demo: switch provider and run the same agent scenario](https://github.com/user-attachments/assets/bf51b8d8-f2ed-464b-bca9-0709fefcc44d)
37
+
36
38
## Why now
37
39
38
40
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.
@@ -45,6 +47,10 @@ OpenAgentPack puts a declarative control plane between the agent and the cloud p
45
47
46
48
The goal is to make an agent an enterprise-controlled, portable, and inheritable digital asset.
47
49
50
+
<palign="center">
51
+
<imgsrc="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">
52
+
</p>
53
+
48
54
### Declaration and portability: the agent as a blueprint
49
55
50
56
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.
@@ -68,18 +74,6 @@ The mechanics are a single `agents.yaml`, a `validate → plan → apply` workfl
68
74
-**Dependency-aware** — Environment → Skill → Agent are created in topological order; a failed dependency skips its dependents instead of leaving half-built state.
69
75
-**Drift recovery** — detects when remote config has drifted from your declaration and reconciles it. The YAML is always the single source of truth.
`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`.
80
-
81
-
[Watch the Playground demo video](https://github.com/user-attachments/assets/bf51b8d8-f2ed-464b-bca9-0709fefcc44d)
Run `agents playground` to launch the local WebUI, and use `--provider` to target `bailian`, `qoder`, `ark`, or `claude`. You can switch providers on the same declaration, run real sessions, and observe tool calls and artifacts.
88
+
89
+
▶ [Watch the full Playground demo](https://github.com/user-attachments/assets/bf51b8d8-f2ed-464b-bca9-0709fefcc44d)
0 commit comments