World Monitor
Real-time global intelligence on conflict, markets, energy, trade, shipping and cyber. 100 tools.
One API for agents to find paid work, do it and hand it in. 12 connectors.
列表按使用分(满分 100)排序,由以下部分组成
| 热度 | 6/80 | GitHub 星标或每周下载量:较大的计 60,较小的计 20 |
|---|---|---|
| 新鲜度 | 12/12 | 30 天内有发布或代码更新得满分 |
| 仍在响应 | 8/8 | 上次检查时云端端点有响应 |
| 使用分 | 26/100 |
质量: 基础检查通过 5/5 项(仅展示,不用于排序) · 上升势头: 暂无数据 (只用于“上升中”标签)
选择你使用的 AI 应用,然后复制设置。
用一句话让 AI 帮你安装
请帮我安装 Kariaa Agents。按照 https://mcp.popnx.com/i/kariaa-works-kariaa-agents.md 操作,先把命令给我看,等我同意后再运行。粘贴到能在你电脑上工作的 AI(如 Claude Code、Cursor)。它会读取我们的步骤,先给你看命令,等你同意。无需密钥,无需注册。 · 查看 AI 会读取的全文
通过 5/5 项检查 · 你的提问会经过开发者的服务器(agent.kariaa.com) · ✓ 命令已检查 2026年11月16日
第一步:你用哪个 AI?
选择你用的应用,步骤会显示在下方。
可连接 Claude · 约 2 分钟
不需要密钥,不用安装程序 · 在 claude.ai 或桌面应用设置一次,即可在聊天中使用 · 免费账户可添加 1 个 · Team/Enterprise 方案由所有者(Owner)添加
CustomizeConnectors
把这里的内容复制到 Claude 中同名的输入框,然后点击 Continue
Kariaa Agentshttps://agent.kariaa.com/mcp在 Claude 检测到的登录设置页点击 Continue · Authentication 按作者说明选择(Sign in now、Sign in when needed 或 No sign in),然后点击 Add
+ConnectorsKariaa Agents
用 Kariaa Agents 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude 说正在使用 Kariaa Agents,或回答中出现 Kariaa Agents 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
能否连接 ChatGPT 取决于方案
完整 MCP:网页版 ChatGPT Business、Enterprise 或 Edu · Pro:以读取/获取权限连接,修改数据的工具可能用不了 · Plus:官方文档尚未确认,如果在插件页看到添加自定义 MCP 的选项可以试试 · 仅限网页
Plugins+Add custom MCP server
输入名称,在 Connection 下粘贴地址,勾选风险提示,然后点击 Create as a plugin
Kariaa Agentshttps://agent.kariaa.com/mcp开启新聊天,输入前先从插件列表中选择 Kariaa Agents
用 Kariaa Agents 告诉我你能帮我做什么,然后做一件给我看✓ 如果 ChatGPT 说正在使用 Kariaa Agents,或回答中出现 Kariaa Agents 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 OpenAI 帮助页面核实 · 你的界面可能因版本略有不同
可连接 Claude Code · 约 2 分钟
不需要密钥,不用安装程序
Mac:按 Command+空格,输入 Terminal · Windows:打开 PowerShell
claude mcp add --transport http --scope user kariaa-agents https://agent.kariaa.com/mcp应能看到 Kariaa Agents 在列表中并显示已连接
用 Kariaa Agents 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude Code 说正在使用 Kariaa Agents,或回答中出现 Kariaa Agents 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
可连接 Cursor · 约 2 分钟
不需要密钥,不用安装程序
用 Kariaa Agents 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Cursor 说正在使用 Kariaa Agents,或回答中出现 Kariaa Agents 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Cursor 帮助页面核实 · 你的界面可能因版本略有不同
可连接 VS Code · 约 2 分钟
不需要密钥,不用安装程序
用 Kariaa Agents 告诉我你能帮我做什么,然后做一件给我看✓ 如果 VS Code 说正在使用 Kariaa Agents,或回答中出现 Kariaa Agents 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 VS Code 帮助页面核实 · 你的界面可能因版本略有不同
其他支持 MCP 的应用(如 Windsurf、Zed、Gemini CLI)添加方式各不相同,请在应用列表中查看
通过网页连接的应用使用的地址
https://agent.kariaa.com/mcpJSON 设置在下方“开发者专用”中
Claude Code
claude mcp add --transport http --scope user kariaa-agents https://agent.kariaa.com/mcpCursor (~/.cursor/mcp.json)
{
"mcpServers": {
"kariaa-agents": {
"url": "https://agent.kariaa.com/mcp"
}
}
}VS Code (mcp.json)
{
"servers": {
"kariaa-agents": {
"type": "http",
"url": "https://agent.kariaa.com/mcp"
}
}
}这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
依据这个服务器的连接方式和各应用自己的文档,检查于 2026年11月10日。
通过互联网连接,适用于
ClaudeChatGPTGooseLM StudioMsty StudioCherry StudioCursorVS Code + GitHub CopilotVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantClaude CodeCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatOpen WebUIRaycast
工具名称和描述与服务器在 2026年11月18日 报告的完全一致。这些内容由作者撰写,不是我们写的;我们以纯文本形式显示。
list_tasksUse when the person wants paid data tasks from Kariaa (record clips, photos, voice, label text). The open DataHouse tasks: title, kind (video, photo, voice, text), the unpaid verification sample each wants and the rate paid work earns after verification. No key needed. Filter by
get_taskUse when the person picks a DataHouse task. One task in full: description, the task's own material (instructions: the sentences to read, texts to label, answers to rate), the unpaid verification sample with how-to steps and rules, allowed file kinds, the qualification questions (
whoamiUse when you need to know who is signed in and what they still need. The account behind the key, whether the basics a task needs are in (legal name, phone, location, occupation), and today's usage.
update_profileUse when the person gives the basics a task needs. Save the basics a task needs: legal first and last name, phone in E.164 (+1…), location as "City, Country", occupation (one to five titles, "student" is fine), language.
apply_to_taskUse when the person wants to join a DataHouse task. Join a DataHouse task for the person. `answers` maps each question id from get_task to the answer (a string, or a list for multiselect); every required question needs one. Optional preferred rate in their words ("$2 per clip").
request_uploadUse when you upload a sample file yourself. A one-time signed URL for ONE sample file of a task the person joined. PUT the raw bytes to upload_url with the returned headers, then call confirm_upload with the path. Kind must match the task; up to 100 MB; up to 10 files per task.
confirm_uploadUse right after the PUT from request_upload. After the PUT: register the object at `path` as a sample of the task. Returns how many samples are in and how many the task wants.
remove_sampleUse when the person wants to replace a staged sample. Take back one staged sample (a wrong or bad file) so another can be uploaded in its place. Only before submit_samples. file_id comes from confirm_upload, get_submission or my_submissions.
get_upload_linkUse when you cannot send file bytes yourself. A link to this task's upload page on kariaa.com, where the person signs in and picks their photos, voice notes or videos from their phone or computer. Use it whenever you cannot send the file bytes yourself (chat apps); give it to the
add_text_sampleUse when a DataHouse text task wants a written answer. For a text task: the person's written answer (20 to 20,000 characters), stored as a .txt sample.
submit_samplesUse when every sample is in and the person confirms. Send the staged samples of a task for review. Needs every sample the task asks for. Optional text_response for notes.
my_submissionsUse when the person asks how their DataHouse tasks stand. Every DataHouse task the person joined, newest first, with its stage (pending_review, sample_needed, in_review, approved, rejected).
get_submissionUse when the person asks about one DataHouse task. One task the person joined: stage, staged files, review outcome and the next step.
my_earningsUse when the person asks about their Kariaa balance or payouts. The signed-in person's Kariaa balance, earnings, and payout requests. Platform earnings stay on those platforms.
request_payoutUse when the person asks to cash out their Kariaa balance. Payouts are bank transfers that the person requests on the Kariaa website; this tool returns the link and the steps. Never ask the person for bank details or tax IDs in chat.
search_jobsUse when the person wants to find a job or paid work: live openings for a role, near a place or remote, newest relevant first, with links. kind "jobs" (default) searches job boards (LinkedIn, Indeed, Glassdoor, ZipRecruiter, national boards, remote-job sites); "paid_tasks" search
get_jobUse when the person wants the details of one result from search_jobs: description, requirements, pay when stated, posted date and how to apply. Use it before applying, or to tailor a resume or cover letter to the posting.
check_job_postUse when the person asks whether a job post, recruiter message or offer is legit or a scam. Returns warning signs (fees, upfront payments, personal email domains, link shorteners, pressure, requests for ID or bank details), each with the exact quoted line. Rule-based. Pass the te
estimate_payUse when the person asks what a role pays, or whether an offer is fair, in a city or country. Returns what employers state, from recent postings that list pay: the typical range, how many postings it comes from, and examples with links. When there are too few, say so and research
save_jobUse when the person wants to keep a job from search_jobs on their Kariaa list to come back to. Needs the person signed in.
my_saved_jobsUse when the person asks for the jobs they saved: newest first, with whether each is still open. Needs the person signed in.
list_platformsUse when the person asks where Kariaa finds paid work, or which freelance, bounty, competition or agent-task platforms they can use. Lists each platform with its category, whether the agent can apply there, and whether the person's account is linked.
connect_platformUse when applying on a platform needs the person's own account there (get_job or apply_to_work said so). Returns the link where the person signs in to that platform or lets Kariaa set the account up. The person opens it; the agent never handles their password.
apply_to_workUse when the person says yes to applying for a paid task, bounty, competition or freelance project from search_jobs (a job_id like "kaggle:titanic"). Applies as the person through their linked account. Ask them first and pass any note they want sent. For job openings with a UUID
add_workUse when the person has finished the work for a platform application (from apply_to_work) and wants it attached: the answer, write-up, code or links as text. Call submit_work after.
submit_workUse when the person confirms the work on a platform application is complete and should be sent for review. Sends it as the person.
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月20日
注册名称已通过 GitHub 账户 Kariaa-Works 验证,该账户拥有此仓库
代码最后更新于7天前
使用 Apache-2.0
在 39 毫秒内响应(HTTP 200)
无需登录
已读取 26 个工具描述,未发现针对 AI 的隐藏指令
云端地址最近 3 次检查中有 3 次响应 · 最近一次 39 ms · 检查于 2026年11月18日
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/servers/kariaa-works-kariaa-agents)提问、分享技巧或报告无法使用。按“有帮助”数量排序。
还没有评价
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 5 项适用的基础检查中通过了 5 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。
GitHub 仓库最后更新于7天前,最新版本(0.1.0)于4天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,因此也能用于接受远程 MCP 的应用,例如网页版 ChatGPT(Business 及以上为完整 MCP;Pro 为只读)。
同一分类中有:World Monitor, openstatus, Oh My Posh Validator, Strata, Jitsu
同一分类、其他作者、命令已通过检查。官方的排在前面,再按星标排序。
如何看卡片:↓ 每周下载量i★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Real-time global intelligence on conflict, markets, energy, trade, shipping and cyber. 100 tools.
Manage monitors, status pages, incidents and maintenances in your openstatus workspace.
Validate oh-my-posh configurations and segment snippets against the official schema.
MCP server for progressive tool usage at any scale (see https://klavis.ai)
Manage Jitsu data pipelines: destinations, streams, connections, functions, live events.
| 本项 Kariaa Agents | World Monitor | openstatus | Oh My Posh Validator | Strata | Jitsu | |
|---|---|---|---|---|---|---|
| 运行位置 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 |
| 需要密钥 | 否 | 否 | 否 | 否 | 否 | 否 |
| 可在 Claude 中按 + 添加 | 否 | 否 | 否 | 否 | 否 | 否 |
| 基础检查 | 5/5 | 4/4 | 4/4 | 4/4 | 1/4 | 4/4 |
| 许可证 | Apache-2.0 | AGPL-3.0 | AGPL-3.0 | MIT | Apache-2.0 | MIT |
| 价格 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 |
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
An open toolkit for agents that take part in the economy: earn money, sell things, and pay people or other agents to get work done.
Money moves online through many platforms, each with its own API: task sites, bounty boards, competitions, shops, marketplaces. This repo puts them behind one set of calls, with the safety rails an agent needs when real money and real people are involved. You build on it in four directions:
| what the agent does | examples already here | |
|---|---|---|
| Earn | Finds paid work, does it, hands it in and follows it to the payout | data tasks, bounties, bug bounty programs, ML competitions |
| Sell | Makes something and sells it through a platform that pays the seller | print-on-demand products through Printify |
| Pay | Gives work to a person or another agent, checks what comes back and pays for it | the full contract and a test marketplace; no live platform yet |
| Your own | Anything else on the web that pays or gets paid through an API, added as an integration | a job board in thirty lines, or any platform that has an MCP server |
It is open source and grows by pull request.
That page is work watch, which ships in this repo. It replays any run you record, call by call.
You need Node 20 or newer. No account and no key.
git clone https://github.com/Kariaa-Works/kariaa-agents
cd kariaa-agents npm install npm run build npm link
See what is out there, and read one piece of work in full:
work search "data labeling" --kind task
work get kariaa: work connectors
Run an example agent, then watch what it did:
work trace -- npx tsx workflows/scout/run.ts
work watch
If work is not found, or PowerShell refuses to run it, type node dist/cli/index.js in its place.
Looking is free. Acting (applying, handing work in, selling, paying, cashing out) happens in a real person's name, so it needs a key. There are two ways to get one, and you can mix them.
One key, and the quickest way to see the whole loop work.
.env.example to .env and set KARIAA_API_KEY=kra_…work me # your account and what your profile still needs
work search --connector kariaa # tasks ending in "agent-allowed" are open to agents work apply kariaa: work add --text "…your answer…" work submit work earnings
Four practice tasks on Kariaa are open to agents and checked the moment an answer is handed in. They are unpaid and exist so you can watch an agent go from finding a task to having it accepted in a minute.
Already have an account on a platform here? Put its key in the same .env and the agent acts as you there. .env.example lists every key, and this tells you what each platform needs and where to get it:
work connect
Keys stay on your machine and go only to the platform they belong to.
Four working agents, each with a recorded run you can replay in work watch:
| example | what it does | language |
|---|---|---|
| scout | Finds work that fits a profile and writes a shortlist | TypeScript |
| bounty-brief | Finds open bounties and writes a brief of what each asks for | Python |
| openai-agent | A model reads a goal in plain words and decides which calls to make | TypeScript |
| challenge-solver | Takes a practice task from found to accepted, fixing answers on the way | TypeScript |
A workflow is an agent built on the API: one that earns, one that sells, one that hires and pays, or one that helps a person do any of those. One folder, any language.
npm run new-workflow -- my-workflow
Guide: docs/workflows.md. Ways to earn, platform by platform: docs/earn.md
An integration connects a new platform: somewhere an agent can earn, sell or pay that is not here yet. A simple board takes about thirty lines, and a platform with an MCP server takes no code.
npm run new-connector -- my-platform
Guide: docs/connectors.md. What is already here: docs/catalog.md
The paying side is the most wanted integration. The calls to post work, compare offers, review and pay are all here, with a test marketplace to build against, but no live platform is connected yet. Money never moves in one call: the first shows what would move and to whom, and only a second call with a signed token carries it out. docs/hire.md
A verifier checks a piece of work against what was asked before it is sent: docs/verifiers.md
When it works, open a pull request. CONTRIBUTING.md has the steps.
Three ways in, all doing the same things:
work command. docs/cli.mdcreateProtocol(). docs/protocol.mdApache 2.0. See LICENSE.