🥒 Pickle: the free AI ops manager for ClickUp, Slack & Teams

ClickUp Brain reads your tasks. Slack AI reads your messages. Both are paid add-ons, and each only sees its own app.
Power your agent
คำสั่งของผู้เผยแพร่เอง ไม่ใช่เอนด์พอยต์ของ PlugMCP
Host a copy with our keys and logs.
npx -y [email protected]Claude Code ไม่มีลิงก์ติดตั้งสำหรับคำสั่งแบบโลคัล ให้คัดลอกคำสั่งด้านล่างแล้วรันในเทอร์มินัล
CLICKUP_API_KEY=<secret> SLACK_TOKEN=<secret> TEAMS_TOKEN=<secret> claude mcp add io.github.adityaarsharma-pickle -- npx -y pickle-mcp@1.1.0รันในเทอร์มินัล
{ "mcpServers": { "io.github.adityaarsharma-pickle": { "command": "npx", "args": [ "-y", "[email protected]" ], "env": { "CLICKUP_API_KEY": "<secret>", "SLACK_TOKEN": "<secret>", "TEAMS_TOKEN": "<secret>" } } }}claude_desktop_config.json
URL แบบรีโมตหรือคำสั่งบนเครื่องของคุณ? การติดตั้งสองแบบต่างกันอย่างไร
รันบน PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
เปิดจากผู้เผยแพร่

ClickUp Brain reads your tasks. Slack AI reads your messages. Both are paid add-ons, and each only sees its own app.
Free, local MCP server that audits ClickUp, Slack & Teams for work that slipped through.
Pickle is a free, open-source AI ops manager for your team, packaged as an MCP server you run yourself. Give it your own ClickUp / Slack / Teams token, and your AI client (Claude, Cursor, Codex, Cline, Zed…) audits your workspace the way a sharp ops manager would: catching the work that quietly fell through.
No account. No hosted server. No key from us. No telemetry. Pickle adds no middleman. It talks only to your tools, with your token. Nothing is sent to us, because there's no "us" to send it to.
⭐ Pickle is free and will stay free. There is no paid tier. If you think workspace tools should run on your own machine, star the repo. Stars are the only way it gets found.
Honest version of the privacy claim: Pickle does read your ClickUp/Slack/Teams. That's the job. But it reads them directly, from your machine, with your token, and sends nothing to any Pickle server. Verify it yourself. The code is in server-remote/server.mjs, and there is no analytics, no beacon, no phone-home.
Guided: it asks which tools you use, points you to where each token lives, and writes your MCP config for you.
curl -fsSL https://pickle.adityaarsharma.com/install.sh | bash
Or add it to your MCP client yourself (via npm):
{
"mcpServers": {
"pickle": {
"command": "npx",
"args": ["-y", "pickle-mcp"],
"env": { "CLICKUP_API_KEY": "pk_your_own_token" }
}
}
}
Add SLACK_TOKEN (xoxp-…) and/or TEAMS_TOKEN to the same env block to switch on the cross-tool patterns. Restart your client. npx runs it locally on your machine, no global install, no clone.
The installer asks which tools you use (ClickUp / Slack / Teams), takes the token you already have for each, and writes the MCP config for your client automatically. No hand-editing, no key from us.
Prefer to do it by hand? See
docs/manual-install.md. The only token you ever provide is your own platform token (e.g. ClickUppk_…from Settings → Apps).
Then just ask your AI:
"Pickle, run my morning audit: scan my ClickUp from the last 7 days and show me the 3 worst things I missed, worst first."
docs/patterns.mdClaude Code · Claude Desktop · Cursor · Codex · Cline · Continue · Zed · any MCP host. The reasoning runs in your model. Pickle just fetches and structures the data. Optional Claude Code skills add /pickle-* slash commands as a convenience layer.
Pickle has no paid tier and never will. Instead:
Pickle is what I use to keep my own team honest. If you're rolling AI into how your company works (onboarding people onto it, wiring it into real workflows, choosing which LLM for what), reach out to me. I help teams go from "we bought AI licenses" to "AI is actually in our workflow," with real use-cases instead of hype.
Built in the open by Aditya Sharma, a marketer who codes.
Is Pickle really free? Yes. MIT-licensed, no paid tier, no account, no credit card. I built it in the open; a ⭐ is the only thing I ask.
Do I need a Pickle account or an API key from you? No. Pickle issues nothing. The only token you provide is your own ClickUp / Slack / Teams token, and it lives in your local MCP config.
Does my data get sent to you or to any cloud?
No middleman. Pickle runs on your machine and talks only to your tools (ClickUp / Slack / Microsoft Graph) using your token. There is no Pickle server, no telemetry, no phone-home. Pickle does read your ClickUp/Slack/Teams (that's the job), but it sends nothing to us. Verify it in server-remote/server.mjs.
Which AI clients does it work with? Any MCP-compatible host: Claude Code, Claude Desktop, Cursor, Codex, Cline, Continue, Zed, and more. The reasoning runs in your model.
What should I ask Pickle first? After installing, restart your AI client and paste: "Pickle, run my morning audit: scan my ClickUp from the last 7 days and show me the 3 worst things I missed, worst first." You get a ranked, punchy result instead of a data dump, so you feel the value on the first run.
How is this different from ClickUp Brain or Slack AI? Those are cloud add-ons that each see only their own tool and send your data to their servers. Pickle runs locally, reads across ClickUp + Slack + Teams, and keeps your token on your machine. Different axis: privacy + cross-tool, not "more features."
Is it safe to give it my token?
The token stays in your local config file (chmod 600), and the code is open. Read it before you connect. Nothing is transmitted anywhere except the platform's own API.
Will Pickle change or delete anything in my tools?
The audit itself is read-only — it only reads and reports. Pickle can create a reminder, drop a comment, or make a task if you explicitly ask it to (for example "remind me about XYZ-184"), because the server also exposes write tools for that. If you want a hard, code-level guarantee that nothing can be written or deleted no matter what a model asks, set PICKLE_READONLY=1 in the env block — it blocks every mutating tool at the server. Details in docs/security.md. You stay in control.
Do I need Slack and Teams too, or is ClickUp enough? ClickUp alone is a great start. Slack/Teams unlock the cross-tool patterns (decisions-in-DMs, ghost mode) because those need chat data.
Can it audit GitHub or other tools? Not yet. A GitHub audit (stale PRs, dropped reviews) is a planned/welcome contribution.
Is there a hosted version? No. Pickle is local by design. If you want your team to adopt AI tools like this well, reach out. That's the consulting I do.
Issues and PRs welcome: new patterns, new connectors (a GitHub audit for stale PRs is a great first one), better client configs.
MIT © Aditya Sharma
Free, local MCP server that audits ClickUp, Slack & Teams for work that slipped through.
{ "mcpServers": { "io.github.adityaarsharma-pickle": { "command": "npx", "args": [ "-y", "[email protected]" ], "env": { "CLICKUP_API_KEY": "<secret>", "SLACK_TOKEN": "<secret>", "TEAMS_TOKEN": "<secret>" } } }}เครื่องมือ
ยังไม่มีรายการเครื่องมือ เซิร์ฟเวอร์แบบรีโมตจะได้รายการหลังการตรวจสอบสถานะครั้งแรก
FAQ
คัดลอกคอนฟิก stdio สำหรับ Cursor, Claude Desktop, Claude Code หรือ VS Code ได้ในหน้านี้ คอนฟิกนี้ใช้การติดตั้งของผู้เผยแพร่ ไม่ใช่พร็อกซีของ PlugMCP
ผู้เผยแพร่ระบุตัวแปรสภาพแวดล้อมเหล่านี้: CLICKUP_API_KEY, SLACK_TOKEN, TEAMS_TOKEN
Pickle ทำงานแบบโลคัลผ่าน stdio
ตรวจสอบล่าสุดเมื่อ 2026-11-09 Free, local MCP server that audits ClickUp, Slack & Teams for work that slipped through.