Merge pull request #177 from modelstudioai/feat/cma-command

Feat/cma command 补齐managed agent cli能力

Author: chenanran555

CHANGELOG.md

@@ -6,6 +6,29 @@ The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and
 
 [中文版](CHANGELOG.zh.md) · [README](README.md) · [Contributing](CONTRIBUTING.md)
 
+## [1.20.0] - 2026-09-03
+
+> Managed Agents now combines YAML-first infrastructure management with direct Bailian AgentStudio resource and runtime operations.
+
+### Added
+
+- **Managed Agent API commands** — added direct list, get, search, version, upload, download, run, pause, archive, event, and diagnostic operations for Agents, Environments, Skills, Vaults, Deployments, Sessions, and Files.
+- **Scoped YAML-backed resource creation** — `agent create`, `environment create`, `skill create`, `vault create`, `vault credential create`, and `deployment create` update `agents.yaml` and apply only the target resource without unrelated drift blocking the operation.
+- **Agent Skill attachment** — Agent creation supports existing custom or official Skill IDs as well as local Skill directories and ZIP archives.
+
+### Changed
+
+- **Bailian-only Managed Agent CLI** — `bl managed-agent` now targets the Bailian provider exclusively; provider-selection flags were removed and configurations containing other providers are rejected.
+- **Runtime mutation confirmation** — Deployment run/pause/unpause, Session archive/delete, and File delete operations require explicit high-risk confirmation.
+
+### Fixed
+
+- **Managed Agent error reporting** — Apply and scoped-create failures preserve the underlying provider diagnostic instead of ending with only `Apply failed.`.
+
+### Security
+
+- Credentials are resolved in memory and removed from the process environment; Vault credential declarations reference environment variables without persisting plaintext secrets.
+
 ## [1.19.0] - 2026-09-01
 
 ### Added

CHANGELOG.zh.md

@@ -6,6 +6,29 @@
 
 [English](CHANGELOG.md) · [README](README.zh.md) · [参与贡献](CONTRIBUTING.zh.md)
 
+## [1.20.0] - 2026-09-03
+
+> Managed Agent 现在同时提供 YAML-first 基础设施管理与百炼 AgentStudio 资源、运行时 API 操作。
+
+### 新增
+
+- **Managed Agent API 命令** —— 新增 Agent、Environment、Skill、Vault、Deployment、Session 和 File 的列表、详情、搜索、版本、上传、下载、运行、暂停、归档、事件及诊断等操作。
+- **基于 YAML 的单资源创建** —— `agent create`、`environment create`、`skill create`、`vault create`、`vault credential create` 和 `deployment create` 会更新 `agents.yaml`，并且只 Apply 目标资源，不受无关资源 Drift 阻塞。
+- **Agent Skill 挂载** —— 创建 Agent 时支持引用已有的自定义或官方 Skill ID，也支持本地 Skill 目录和 ZIP 文件。
+
+### 变更
+
+- **Managed Agent CLI 限定为百炼 Provider** —— `bl managed-agent` 现在只面向百炼，移除 Provider 选择参数，并拒绝包含其他 Provider 的配置。
+- **运行时变更增加确认** —— Deployment 运行/暂停/恢复、Session 归档/删除以及 File 删除操作需要显式进行高风险确认。
+
+### 修复
+
+- **Managed Agent 错误输出** —— Apply 和单资源创建失败时会保留底层 Provider 的具体诊断，不再只显示 `Apply failed.`。
+
+### 安全
+
+- 凭证仅在内存中解析并从进程环境清除；Vault Credential 声明通过环境变量引用 Secret，不会持久化明文。
+
 ## [1.19.0] - 2026-09-01
 
 ### 新增

docs/agents/auth-change.md

@@ -62,17 +62,19 @@ defineCommand({ auth }) → runtime/authStage → ctx.client → command.run(ctx
 `bl managed-agent *` 按调用链分两层:
 
 - **离线命令** — `init`、`validate`、`state list/show/rm`:`auth: "none"`，只读写本地文件，无需登录;引擎侧传 `credentials: "none"` 跳过凭证断言
-- **联网命令** — `plan`、`apply`、`destroy`、`state import`、`skill-list`、全部 `session *`:统一声明 `auth: "apiKey"` 硬门禁 —— 无论目标 provider 是谁，authStage 都经 `resolveApiKey(sources)` 解析 bailian 凭证(flag > env > active profile config)，缺失报统一 AUTH;引擎层 `assertProviderCredentials` 再对 agents.yaml 里**全部已声明 provider** 的空 key 拦截并给 provider 专属 hint。例外:`plan --no-refresh` / `plan --dry-run` 传 `credentials: "none"` 并强制 `refresh: false`（不联网、不回写 state，不查 provider key），其中 `--dry-run` 连登录也不要求（authStage 的 dry-run 豁免），`--no-refresh` 仍需登录。
+- **联网命令** — `plan`、`apply`、`destroy`、`state import`、`skill-list`、全部 `session *`:统一声明 `auth: "apiKey"` 硬门禁，authStage 经 `resolveApiKey(sources)` 解析 Bailian 凭证(flag > env > active profile config)，缺失报统一 AUTH;引擎层再断言 Bailian key 非空。例外:`plan --no-refresh` / `plan --dry-run` 传 `credentials: "none"` 并强制 `refresh: false`（不联网、不回写 state，不查 provider key），其中 `--dry-run` 连登录也不要求（authStage 的 dry-run 豁免），`--no-refresh` 仍需登录。
+
+`bl managed-agent` 是 Bailian-only 产品入口：命令不暴露 `--provider`，`init` 只生成 `providers.bailian`，所有远端调用固定传 `provider: "bailian"`。`resolveAgentProjectConfig` 在创建 SDK runtime 前通过 `assertBailianOnlyProviders` 拒绝包含非 Bailian Provider 的手写配置；共享 `@openagentpack/sdk` 仍可保留多 Provider 能力。
 
 凭证不以真实值写入 `process.env`，而是经 `packages/commands/src/commands/managed-agent/_engine/` 的**内存注入管道**(`resolveAgentProjectConfig`)注入 SDK，管道五步:
 
-1. `prepareProviderEnv()` — 先 `bootstrapRuntimeCredentialsSync()`(SDK 把 `.env` / `~/.agents/config.json` 灌进 env，服务 claude/ark/qoder 等非 bailian provider)，再把全部凭证类 env(`CREDENTIAL_ENV_KEYS`，含别名)中仍为 undefined 的占位为 `""`，使 agents.yaml 插值不因缺变量抛错
-2. `resolveProjectConfig` — 插值发生:bailian 插值拿到占位空串，claude/ark 拿到真实 env 值;随后 `normalizeInterpolatedProviderBlocks()` 把插值为空导致的 YAML `null` 归一为 `""`(避免离线命令下空 key 在 SDK zod 层报 "received null")
+1. `prepareProviderEnv()` — 调用 SDK 的凭证 bootstrap，再把凭证类 env(`CREDENTIAL_ENV_KEYS`，含兼容别名)中仍为 undefined 的项占位为 `""`，使 agents.yaml 插值阶段能够完成并由 CLI 输出明确的 Bailian-only 配置错误
+2. `resolveProjectConfig` — 完成插值；随后 `normalizeInterpolatedProviderBlocks()` 把插值为空导致的 YAML `null` 归一为 `""`，避免空 key 在 SDK zod 层提前报 "received null"
 3. `injectProviderCredentials()` — 用 `ctx.client.exportApiCredential()`(lint 限定 `managed-agent/_engine/**` 可用)覆写内存 config 对象的 bailian 块:有凭证时 `api_key` 无条件覆写;`base_url`(拼 `/api/v1/agentstudio` 后缀，无凭证时用 client 默认域名补齐以满足 schema)/`workspace_id`(取 `settings.workspaceId`)仅在引用且为空时填充
 4. `scrubCredentialEnv()` — 从 `process.env` 删除全部凭证变量(真实凭证此后只存于 config 对象 → provider adapter 实例内存，不驻留 env / 不被子进程继承)
-5. `assertProviderCredentials(providers)` — 任一已声明 provider 的 `api_key` 为空 → CLI 权威 `AUTH` 错误 + provider 专属 hint(取代 SDK 原始插值/zod 报错);离线命令传 `credentials: "none"` 整体跳过
+5. `assertBailianOnlyProviders(providers)` — 拒绝非 Bailian Provider；随后 `assertProviderCredentials(providers)` 在 Bailian `api_key` 为空时给出 CLI 权威 `AUTH` 错误和登录 hint；离线命令传 `credentials: "none"` 跳过 key 断言，但仍执行 Bailian-only 配置校验
 
-`bl auth login` 仅管理 bailian(DashScope)凭证;claude/ark/qoder 的 key 从 env(shell / `.env` / `~/.agents/config.json`)经插值进入 config 对象，同样被清扫。禁止命令层直接 `readConfigFile` 裸读凭证;bailian 字段以 CLI 鉴权链为唯一信源。
+禁止命令层直接 `readConfigFile` 裸读凭证；Bailian 字段以 CLI 鉴权链为唯一信源。SDK bootstrap 期间读取到的兼容凭证变量也会在配置解析后统一清扫。
 
 ## 必查清单
 

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "bailian-cli",
-  "version": "1.19.0",
+  "version": "1.20.0",
   "description": "CLI for Aliyun Model Studio (DashScope) AI Platform.",
   "keywords": [
     "agent",

packages/cli/src/commands.ts

@@ -152,6 +152,48 @@ import {
   managedAgentSessionSend,
   managedAgentSessionEvents,
   managedAgentSkillList,
+  managedAgentAgentCreate,
+  managedAgentEnvironmentCreate,
+  managedAgentSkillCreate,
+  managedAgentVaultCreate,
+  managedAgentVaultCredentialCreate,
+  managedAgentDeploymentCreate,
+  managedAgentAgentList,
+  managedAgentAgentGet,
+  managedAgentAgentSearch,
+  managedAgentAgentVersions,
+  managedAgentEnvironmentList,
+  managedAgentEnvironmentGet,
+  managedAgentEnvironmentSearch,
+  managedAgentSkillGet,
+  managedAgentSkillSearch,
+  managedAgentSkillVersions,
+  managedAgentSkillDownload,
+  managedAgentVaultList,
+  managedAgentVaultGet,
+  managedAgentVaultSearch,
+  managedAgentDeploymentList,
+  managedAgentDeploymentGet,
+  managedAgentDeploymentSearch,
+  managedAgentDeploymentRunsList,
+  managedAgentDeploymentRunsGet,
+  managedAgentDeploymentRun,
+  managedAgentDeploymentPause,
+  managedAgentDeploymentUnpause,
+  managedAgentSessionSearch,
+  managedAgentSessionUpdate,
+  managedAgentSessionArchive,
+  managedAgentSessionEventList,
+  managedAgentSessionEventSend,
+  managedAgentSessionEventStream,
+  managedAgentSessionDebug,
+  managedAgentSessionExport,
+  managedAgentFileUpload,
+  managedAgentFileList,
+  managedAgentFileGet,
+  managedAgentFileSearch,
+  managedAgentFileDownload,
+  managedAgentFileDelete,
 } from "bailian-cli-commands";
 
 // Full bailian-cli product: every command, exposed under the `bl` binary.
@@ -314,6 +356,49 @@ export const commands: Record<string, AnyCommand> = {
   "managed-agent session send": managedAgentSessionSend,
   "managed-agent session events": managedAgentSessionEvents,
   "managed-agent skill-list": managedAgentSkillList,
+  "managed-agent agent create": managedAgentAgentCreate,
+  "managed-agent agent list": managedAgentAgentList,
+  "managed-agent agent get": managedAgentAgentGet,
+  "managed-agent agent search": managedAgentAgentSearch,
+  "managed-agent agent versions": managedAgentAgentVersions,
+  "managed-agent environment create": managedAgentEnvironmentCreate,
+  "managed-agent environment list": managedAgentEnvironmentList,
+  "managed-agent environment get": managedAgentEnvironmentGet,
+  "managed-agent environment search": managedAgentEnvironmentSearch,
+  "managed-agent skill create": managedAgentSkillCreate,
+  "managed-agent skill list": managedAgentSkillList,
+  "managed-agent skill get": managedAgentSkillGet,
+  "managed-agent skill search": managedAgentSkillSearch,
+  "managed-agent skill versions": managedAgentSkillVersions,
+  "managed-agent skill download": managedAgentSkillDownload,
+  "managed-agent vault create": managedAgentVaultCreate,
+  "managed-agent vault credential create": managedAgentVaultCredentialCreate,
+  "managed-agent vault list": managedAgentVaultList,
+  "managed-agent vault get": managedAgentVaultGet,
+  "managed-agent vault search": managedAgentVaultSearch,
+  "managed-agent deployment create": managedAgentDeploymentCreate,
+  "managed-agent deployment list": managedAgentDeploymentList,
+  "managed-agent deployment get": managedAgentDeploymentGet,
+  "managed-agent deployment search": managedAgentDeploymentSearch,
+  "managed-agent deployment runs list": managedAgentDeploymentRunsList,
+  "managed-agent deployment runs get": managedAgentDeploymentRunsGet,
+  "managed-agent deployment run": managedAgentDeploymentRun,
+  "managed-agent deployment pause": managedAgentDeploymentPause,
+  "managed-agent deployment unpause": managedAgentDeploymentUnpause,
+  "managed-agent session search": managedAgentSessionSearch,
+  "managed-agent session update": managedAgentSessionUpdate,
+  "managed-agent session archive": managedAgentSessionArchive,
+  "managed-agent session event list": managedAgentSessionEventList,
+  "managed-agent session event send": managedAgentSessionEventSend,
+  "managed-agent session event stream": managedAgentSessionEventStream,
+  "managed-agent session debug": managedAgentSessionDebug,
+  "managed-agent session export": managedAgentSessionExport,
+  "managed-agent file upload": managedAgentFileUpload,
+  "managed-agent file list": managedAgentFileList,
+  "managed-agent file get": managedAgentFileGet,
+  "managed-agent file search": managedAgentFileSearch,
+  "managed-agent file download": managedAgentFileDownload,
+  "managed-agent file delete": managedAgentFileDelete,
 };
 
 /**

packages/cli/tests/skill-risk-confirmation.test.ts

@@ -5,9 +5,18 @@ import { expect, test } from "vite-plus/test";
 
 const repositoryRoot = join(dirname(fileURLToPath(import.meta.url)), "../../..");
 const skillsRoot = join(repositoryRoot, "skills");
+const scopedCreateCommands = new Set([
+  "bl managed-agent agent create",
+  "bl managed-agent deployment create",
+  "bl managed-agent environment create",
+  "bl managed-agent skill create",
+  "bl managed-agent vault create",
+  "bl managed-agent vault credential create",
+]);
 
-test("every generated high-risk command reference requires user confirmation before --yes", () => {
+test("generated references distinguish runtime high-risk confirmation from scoped create execution", () => {
   let highRiskCommandCount = 0;
+  const seenScopedCreateCommands = new Set<string>();
 
   for (const skillDirectory of readdirSync(skillsRoot, { withFileTypes: true })) {
     if (!skillDirectory.isDirectory()) continue;
@@ -27,9 +36,21 @@ test("every generated high-risk command reference requires user confirmation bef
       const commandSections = markdown.split(/(?=^### `bl )/m).slice(1);
 
       for (const commandSection of commandSections) {
-        if (!commandSection.includes("`--yes`")) continue;
+        const commandName = commandSection.match(/^### `([^`]+)`/m)?.[1];
+        const hasConfirmationFlag = commandSection.includes("`--yes`");
+        const hasHighRiskMetadata = /\|\s+\*\*Risk\*\*\s+\|\s+`high`\s+\|/.test(commandSection);
+
+        if (!hasHighRiskMetadata) {
+          if (!hasConfirmationFlag) continue;
+          expect(commandName).toBeDefined();
+          expect(scopedCreateCommands.has(commandName ?? "")).toBe(true);
+          expect(commandSection).toMatch(/Without --yes, .*preview/i);
+          seenScopedCreateCommands.add(commandName ?? "");
+          continue;
+        }
+
         highRiskCommandCount += 1;
-        expect(commandSection).toMatch(/\|\s+\*\*Risk\*\*\s+\|\s+`high`\s+\|/);
+        expect(hasConfirmationFlag).toBe(true);
         expect(commandSection).toMatch(/\|\s+\*\*Risk message\*\*\s+\|\s+.+\|/);
         expect(commandSection).toMatch(/type=.*requires_confirmation/);
         const agentSafetyLine = commandSection
@@ -44,4 +65,5 @@ test("every generated high-risk command reference requires user confirmation bef
   }
 
   expect(highRiskCommandCount).toBeGreaterThan(0);
+  expect([...seenScopedCreateCommands].sort()).toEqual([...scopedCreateCommands].sort());
 });

packages/commands/package.json

@@ -1,6 +1,6 @@
 {
   "name": "bailian-cli-commands",
-  "version": "1.19.0",
+  "version": "1.20.0",
   "description": "Command library for bailian-cli products (knowledge, memory, media, …). See https://www.npmjs.com/package/bailian-cli for usage.",
   "homepage": "https://bailian.console.aliyun.com/cli",
   "bugs": {
@@ -40,7 +40,7 @@
     "check": "vp check"
   },
   "dependencies": {
-    "@openagentpack/sdk": "0.3.2",
+    "@openagentpack/sdk": "0.5.0",
     "bailian-cli-core": "workspace:*",
     "bailian-cli-runtime": "workspace:*",
     "boxen": "catalog:",