Paperclip

Versus Lindy: the match

• Price

  Paperclip

  €0, plus your agents' subscription or API key

  Lindy

  €26/month per user, AI included in the credits

• The AI

  Paperclip

  Your agents: Claude Code, Codex, OpenCode, Hermes, Cursor…

  Lindy

  Included, model of your choice for each task

• The team

  Paperclip

  Several agents, with roles, an org chart and delegation

  Lindy

  One assistant in Slack, with over 40 skills

See 5 more advantages

Lindy ↗

VS	Paperclip	Lindy ↗
Price	€0, plus your agents' subscription or API key	€26/month per user, AI included in the credits
The AI	Your agents: Claude Code, Codex, OpenCode, Hermes, Cursor…	Included, model of your choice for each task
The team	Several agents, with roles, an org chart and delegation	One assistant in Slack, with over 40 skills
Recurring tasks	Scheduled routines, each one leaves a task and its history	Scheduled routines
Your go-ahead	Plan approvals, and hire approvals if you turn them on	Built-in approvals
Apps	Gmail, Drive, Notion, Slack, Shopify, Zapier and any MCP server	Over 1,500 integrations
Where it runs	On your Mac or your server, only calls to the AI and to connected apps leave it	In Lindy's cloud
Code	Open, MIT licence	Closed

How to install it (even from zero)

You need a Mac (or Linux, or Windows with WSL2), Node.js 24 and Claude Code signed in to your subscription. The prompt installs Paperclip from the official repository, keeps it on your computer only, turns off telemetry and sets the guardrails before your first team. Count twenty minutes.

1. 1

   Paste this prompt into Claude Code

   It installs Paperclip from the official repository, connects your subscription and has you create your first team with the right settings:

   Je veux installer Paperclip (l'application open source pour diriger une équipe d'agents IA, dépôt GitHub officiel paperclipai/paperclip, site paperclip.ing) sur mon ordinateur, en local seulement, à la place d'un service payant comme Lindy. Guide-moi pas à pas en me posant des questions et suis le guide d'installation du dépôt (doc/INSTALLING.md). Pars de zéro : vérifie que Node.js 24.11 ou plus récent est installé (node --version). Sinon, installe Node.js 24 LTS ; si j'utilise déjà une autre version de Node pour d'autres projets, installe la 24 à côté (avec nvm ou node@24 de Homebrew) sans remplacer la mienne, l'installation gérée de Paperclip retient le bon Node. Vérifie aussi que python3 répond et que Claude Code est connecté à mon abonnement Claude (ou Codex à mon abonnement ChatGPT, si je préfère). Installe ensuite Paperclip avec le paquet npm officiel publié par ce dépôt, télémétrie coupée dès le départ : DO_NOT_TRACK=1 npx paperclipai install --yes, puis DO_NOT_TRACK=1 paperclipai onboard --yes --install-service. Ce mode garde le tableau de bord sur mon ordinateur seulement, à l'adresse qu'il affiche (en général http://localhost:3100), et le relance à chaque démarrage. Coupe la télémétrie pour de bon : mets telemetry.enabled à false dans ~/.paperclip/instances/default/config.json, puis lance paperclipai service restart. Ouvre le tableau de bord, crée mon organisation et choisis mon abonnement comme source du modèle, pas une clé d'API ; si une variable ANTHROPIC_API_KEY existe sur ma machine, préviens-moi, car elle passerait devant mon abonnement et je paierais à l'usage. Avant la première tâche, pose les garde-fous : dans les réglages de l'organisation, active Require board approval for new hires (sinon un agent peut en embaucher d'autres sans moi). Explique-moi aussi que le dossier de travail de chaque agent n'est pas une cage : un agent Claude agit sans demander la permission, avec les mêmes droits que moi, donc propose-moi de faire tourner Paperclip sur un compte Mac à part si je veux l'isoler. Lance enfin la première tâche avec le Chief of staff : je lui décris mon activité, il me propose une équipe, je valide. Termine en me montrant comment mettre un agent en pause, arrêter le service (paperclipai service stop) et mettre Paperclip à jour (paperclipai update). Explique chaque étape simplement, je débute.
   
   Review the GitHub repository and check whether it is safe to install given my machine and my infrastructure. Ask me as many questions as possible, using widgets, to clarify my request and be as precise as you can.

   Copy
2. 2

   Know what your agents can reach

   A Claude agent started by Paperclip does not ask for permission: it can read, change or delete anything your account can reach, even outside its working folder. Never hand it a password in plain text and, for peace of mind, run Paperclip on a separate Mac account or on a small server.

3. 3

   Keep control of the bill

   With an API key, set a budget per agent: Paperclip pauses it at 100%. With a subscription, those budgets stay at zero and your Claude or ChatGPT quota sets the limit. Start with one or two agents and space out their wake ups.

What you actually do with it

Project health, at a glance

Before installing an open-source tool, one question matters: is the project alive? These scores are computed from public GitHub data. They are not made up.

Scores computed automatically from the GitHub API (last-30-days activity, update freshness, community size, project age). Refreshed daily.

Also worth a look in the library

• 

  OpenBot

  Agents & orchestration€86/mo

  AI agents that each get their own computer: a real browser with their own logins, files and tools. The same idea as xAI's Grok Bot, except this one runs at home and costs nothing.

  View details →

• 

  n8n

  Agents & orchestration€27/mo

  Wire your tools together and let it run: when this happens, do that. Self-hosted, no task counter stopping you mid-month.

  View details →

• 

  Hermes Agent

  Agents & orchestration€23/mo

  An agent that learns from you and grows with use. Wire it to your second brain, run it on a local model, and you have a personal assistant that talks to nobody but you.

  View details →