Token导航 LogoToken导航

Agent Skill · 研究检索

anthropic-managed-agents(Anthropic managed Agent 搜索)

anthropic-managed-agents 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

anthropic-managed-agents 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为调用外部服务;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 71,089。

总安装 196周安装 8GitHub Stars 55下载 63核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:anthropic-managed-agents(Anthropic managed Agent 搜索)
来源仓库:https://github.com/vm0-ai/vm0-skills
仓库路径:skills/anthropic-managed-agents
安装命令:
npx skills add https://github.com/vm0-ai/vm0-skills --skill anthropic-managed-agents
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/vm0-ai/vm0-skills --skill anthropic-managed-agents

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

anthropic-managed-agents 提供 Anthropic 托管 Agent 的创建、管理与故障排查能力。适用于 Codex、Claude、Cursor、Gemini CLI 中配置可复用的 AI 代理(model + prompt + tools)。

支持通过 JSON 定义 Agent 名称、描述与模型参数,便于跨会话复用与版本控制。安装前需设置 ANTHROPIC_MANAGED_AGENTS_TOKEN,注意 API 权限范围与计费策略。若请求失败,可通过 zero doctor 工具检查连接器状态与认证信息。

来源仓库内路径:skills/anthropic-managed-agents

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/vm0-ai/vm0-skills --skill anthropic-managed-ag…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要调用外部服务,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

anthropic-managed-agents 怎么安装?

在终端执行 npx skills add https://github.com/vm0-ai/vm0-skills --skill anthropic-managed-agents。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

anthropic-managed-agents 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

anthropic-managed-agents 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

anthropic-managed-agents 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

anthropic-managed-agents 用的人多吗?

公开数据是累计安装 196 次,近一周安装 8 次,下载 63 次,GitHub 55 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

anthropic-managed-agents 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
anthropic-managed-agents

Anthropic managed Agent 搜索 · 本页

19655调用外部服务未明确未通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Troubleshooting

If requests fail, run zero doctor check-connector --env-name ANTHROPIC_MANAGED_AGENTS_TOKEN or zero doctor check-connector --url https://api.anthropic.com/v1/agents --method POST

How to Use

All examples assume ANTHROPIC_MANAGED_AGENTS_TOKEN is set. The base URL is https://api.anthropic.com.

1. Create an Agent

An agent is a reusable configuration (model + system prompt + toolset). Create it once; use it across many sessions.

Write to /tmp/agent_create.json:

{
  "model": "claude-sonnet-4-6",
  "name": "Code Review Agent",
  "description": "Reviews pull requests and suggests improvements",
  "system": "You are an expert code reviewer. Analyze the code in the mounted repository, identify issues, and provide actionable feedback.",
  "tools": [
    {
      "type": "agent_toolset_20260401"
    }
  ]
}

Then run:

curl -s -X POST "https://api.anthropic.com/v1/agents" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Content-Type: application/json" \
  -d @/tmp/agent_create.json | jq '{id: .id, name: .name}'

Save the returned id (e.g., agent_011CZk...) for use in sessions.

Built-in tools in agent_toolset_20260401: bash, edit, read, write, glob, grep, web_fetch, web_search

Available models: claude-opus-4-6, claude-sonnet-4-6, claude-haiku-4-5

2. List Agents

curl -s "https://api.anthropic.com/v1/agents" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" | jq '.data[] | {id: .id, name: .name, model: .model}'

3. Get an Agent

curl -s "https://api.anthropic.com/v1/agents/<AGENT_ID>" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" | jq '.'

4. Delete an Agent

DELETE is served by an older beta API and requires a different anthropic-beta header than the other endpoints. Use agent-api-2026-03-01 alone — combining it with managed-agents-2026-04-01 is rejected as incompatible.

curl -s -X DELETE "https://api.anthropic.com/v1/agents/<AGENT_ID>" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: agent-api-2026-03-01"

5. Create an Environment

An environment defines the container configuration (network policy, packages) that sessions run in. Create once, reuse across sessions.

Write to /tmp/env_create.json:

{
  "name": "Default Cloud Environment",
  "description": "Standard environment with unrestricted network access",
  "config": {
    "type": "cloud",
    "networking": {
      "type": "unrestricted"
    },
    "packages": {
      "pip": ["requests", "pandas"],
      "npm": ["typescript"]
    }
  }
}

Then run:

curl -s -X POST "https://api.anthropic.com/v1/environments" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Content-Type: application/json" \
  -d @/tmp/env_create.json | jq '{id: .id, name: .name}'

Save the returned id (e.g., env_011...) for session creation.

Network options:

  • {"type": "unrestricted"} — full internet access
  • {"type": "limited", "allow_package_managers": true, "allow_mcp_servers": true, "allowed_hosts": ["api.github.com"]} — restricted outbound

6. Create a Session (Start a Task)

A session runs one task: it mounts resources (repos, files), runs the agent, and produces output. Each session is a separate container.

Write to /tmp/session_create.json:

{
  "agent": "agent_011CZk...",
  "environment_id": "env_011...",
  "title": "Review PR #42",
  "resources": [
    {
      "type": "github_repository",
      "url": "https://github.com/your-org/your-repo",
      "authorization_token": "ghp_xxx",
      "checkout": {
        "type": "branch",
        "name": "feature/your-branch"
      }
    }
  ]
}

Then run:

curl -s -X POST "https://api.anthropic.com/v1/sessions" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Content-Type: application/json" \
  -d @/tmp/session_create.json | jq '{id: .id, status: .status}'

Resource types:

  • github_repository — mounts a GitHub repo; requires url, authorization_token (GitHub PAT), and optional checkout (branch or commit SHA). Defaults to repo's default branch.
  • file — mounts a file uploaded via the Files API; requires file_id.

Checkout options:

  • {"type": "branch", "name": "main"}
  • {"type": "commit", "sha": "abc123..."}

With Vault secrets:

{
  "agent": "agent_011CZk...",
  "environment_id": "env_011...",
  "resources": [...],
  "vault_ids": ["vault_abc..."]
}

7. Stream Session Events

Stream real-time SSE events from a running session. The stream stays open until the session completes.

curl -s "https://api.anthropic.com/v1/sessions/<SESSION_ID>/events/stream" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Accept: text/event-stream"

Events are Server-Sent Events (SSE). Each event has a type field:

  • agent.message — agent text output
  • agent.tool_use — tool invocation (bash command, file read, etc.)
  • agent.tool_result — tool output
  • agent.thinking — extended thinking progress signal
  • agent.mcp_tool_use — MCP tool invocation
  • agent.mcp_tool_result — MCP tool output

Parse agent message text output only:

curl -s "https://api.anthropic.com/v1/sessions/<SESSION_ID>/events/stream" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Accept: text/event-stream" | grep '^data:' | python3 -c "
import sys, json
for line in sys.stdin:
    line = line.strip()
    if line.startswith('data: '):
        try:
            ev = json.loads(line[6:])
            # agent.message events contain content blocks with text
            if ev.get('type') == 'agent.message':
                for block in ev.get('content', []):
                    if block.get('type') == 'text':
                        print(block.get('text', ''), end='', flush=True)
        except:
            pass
"

7b. Send Events to a Session (Follow-up Messages)

While a session is idle (waiting), send follow-up messages or tool results:

curl -s -X POST "https://api.anthropic.com/v1/sessions/<SESSION_ID>/events" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Content-Type: application/json" \
  -d '{
    "events": [
      {
        "type": "user.message",
        "content": [{"type": "text", "text": "Now focus on the security implications."}]
      }
    ]
  }'

7c. List Past Events

Replay all events from a completed or running session:

curl -s "https://api.anthropic.com/v1/sessions/<SESSION_ID>/events" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" | jq '.data[] | {type: .type, id: .id}'

8. Get Session Status

curl -s "https://api.anthropic.com/v1/sessions/<SESSION_ID>" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" | jq '{id: .id, status: .status, title: .title}'

Session statuses: running, idle, terminated

9. List Sessions

# List all sessions for an agent
curl -s "https://api.anthropic.com/v1/sessions?agent_id=<AGENT_ID>&limit=10" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" | jq '.data[] | {id: .id, status: .status, title: .title}'

Supports filters: agent_id, agent_version, created_at[gt/gte/lt/lte], include_archived, order (asc/desc), limit, page.

10. Full End-to-End Example: Code Review Workflow

# Step 1: Create agent (one-time setup)
AGENT_ID=$(curl -s -X POST "https://api.anthropic.com/v1/agents" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "name": "Code Reviewer",
    "system": "Review the code changes in the repository and provide detailed feedback on correctness, style, and potential issues.",
    "tools": [{"type": "agent_toolset_20260401"}]
  }' | jq -r '.id')
echo "Agent: $AGENT_ID"

# Step 2: Create environment (one-time setup)
ENV_ID=$(curl -s -X POST "https://api.anthropic.com/v1/environments" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Content-Type: application/json" \
  -d '{"name": "Default", "config": {"type": "cloud", "networking": {"type": "unrestricted"}}}' | jq -r '.id')
echo "Environment: $ENV_ID"

# Step 3: Start a session for this PR review task
SESSION_ID=$(curl -s -X POST "https://api.anthropic.com/v1/sessions" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Content-Type: application/json" \
  -d "{
    \"agent\": \"$AGENT_ID\",
    \"environment_id\": \"$ENV_ID\",
    \"title\": \"Code Review\",
    \"resources\": [{
      \"type\": \"github_repository\",
      \"url\": \"https://github.com/your-org/your-repo\",
      \"authorization_token\": \"$GITHUB_TOKEN\",
      \"checkout\": {\"type\": \"branch\", \"name\": \"feature/my-feature\"}
    }]
  }" | jq -r '.id')
echo "Session: $SESSION_ID"

# Step 4: Stream the output until complete
curl -s "https://api.anthropic.com/v1/sessions/$SESSION_ID/events/stream" \
  -H "x-api-key: $ANTHROPIC_MANAGED_AGENTS_TOKEN" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: managed-agents-2026-04-01" \
  -H "Accept: text/event-stream" | grep '^data:' | python3 -c "
import sys, json
for line in sys.stdin:
    line = line.strip()
    if line.startswith('data: '):
        try:
            ev = json.loads(line[6:])
            t = ev.get('type', '')
            if t == 'text':
                print(ev.get('text', ''), end='', flush=True)
            elif t == 'status_change':
                print(f\"\n[Status: {ev.get('status')}]\", flush=True)
        except:
            pass
"

Guidelines

  1. Agent lifecycle: Create an agent once; reuse its ID for all sessions. Agents are versioned — each update increments the version.
  2. Environment lifecycle: Create an environment once per container config. Reference the same env_id across many sessions.
  3. Sessions are single-use: Each session runs one task in an isolated container. For follow-up questions, create a new session.
  4. Stream before polling: Use the SSE stream (/events/stream) to get real-time output. Only fall back to GET /v1/sessions/{id} for status checks after the stream closes.
  5. GitHub auth: Pass a GitHub PAT with repo scope as authorization_token in repository resources. The token is used only for that session.
  6. Vaults for secrets: Store long-lived credentials (API keys, tokens) in a Vault and pass vault_ids instead of embedding secrets in session payloads.
  7. Model selection: Use claude-sonnet-4-6 for most coding tasks (speed + quality balance). Use claude-opus-4-6 for complex multi-file refactors.
  8. Beta header required: Most Managed Agents endpoints require anthropic-beta: managed-agents-2026-04-01. The sole exception is DELETE /v1/agents/{id}, which requires agent-api-2026-03-01 instead — the two beta values are mutually exclusive and cannot be combined in one request.
  9. Billing: Sessions consume Claude API credits (tokens), not claude.ai subscription quota. Monitor usage in Claude Console.

平台分布

Codex

32.48%
按下载量换算20

Claude

32%
按下载量换算20

Cursor

17.29%
按下载量换算11

Gemini CLI

9.59%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

可疑

Snyk

未通过

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills