← Back to library

Separate agent bots for your repo, notes and team

A developer runs several OpenClaw-style instances. One sits in Discord and maintains a personal project, logging issues, setting priorities and updating docs. Another takes notes while driving or walking and sends a daily digest of AI news from sources they trust. A third works in the team Slack, sending weekly and monthly meeting recaps and fielding questions from other teams.

  • OutcomeRun a few agent instances with separate jobs, one maintaining a side project, one taking notes and news, and one serving your team, instead of one bot for everything.
  • SetupAn afternoon+
  • LevelAdvanced
  • Creditdevashish86 on Hacker News
View original
Works with
OpenClaw

Published Oct 23, 2026Updated Oct 23, 2026AI dry run on sample input: passed Oct 23, 2026Source verified Oct 23, 2026

When should you use it?

  • One bot doing everything gets confused
  • You want work and personal kept apart
  • You can host several instances

Not for: Single simple tasks; Teams without approval to add a bot

What do you need, and what do you get?

You need

  • Separate chat channels
  • A job description for each bot
  • Access limited per bot

You get

  • Repo upkeep from chat
  • Captured notes plus a morning news roundup
  • Meeting recaps and replies for the team

How does it work, step by step?

  1. Split the jobsWrite a short role for each bot.
  2. Separate accessGive each bot only the accounts its role needs.
  3. Pick channelsPut each bot in the chat where its work happens.
  4. Set outputsDecide what each sends and when, such as weekly summaries.
  5. Manage themUse a simple dashboard or script to track all instances.

The prompt

Our own prompt, inspired by devashish86 on Hacker News and written to work in OpenClaw. Values in lime are placeholders.

Fill in the blanks (optional; the copied prompt uses your values)

Prompt6 variables
You are the {role} bot. Your only jobs: {jobs}. You may use only {accounts}.
Post in {channel}. Send {output} on {schedule}. If asked to do something outside your role, say which bot handles it instead.

Inspired by devashish86 on Hacker News: Comment in Ask HN: Who is using OpenClaw?. The prompt above is written by playbooks.bot in our own words; nothing from the original is quoted.

What can go wrong?

  • Team bots see company data; get approval.
  • Several bots multiply cost; watch usage.
  • Keep personal notes away from the work bot.

Our take

devashish86's setup is a good argument for giving agents narrow jobs instead of one assistant with every key. The repository bot lives in Discord, takes thrown-together ideas and turns them into issues, priorities and doc updates, which is maintenance work that otherwise piles up. The personal bot handles notes captured on walks or drives and a daily news digest filtered by the owner's own relevance rules and trusted sources. The team bot sits in Slack, writes weekly and monthly meeting recaps and answers questions from other teams. Because each has a short brief and its own channel, it is easy to tell which one misbehaved and to fix it without disturbing the others. Separation also protects privacy: the bot your colleagues can talk to has no reason to see your personal notes, so give it no access to them. Before adding any bot to a workplace chat, get approval from whoever runs it and tell the team what it reads and stores. Track combined running costs monthly, since several always-on agents add up quietly.

Where the time goes back: Each bot stays focused, and work data stays apart from personal notes.

FAQ

Which bot was most useful?
The work Slack assistant, which the developer calls the most real use.
How did they manage several bots?
They built a small orchestration tool to avoid manual server setup.

Source and credit

Inspired by devashish86 on Hacker News. We describe the idea in our own words, link to the original, and wrote the prompt ourselves; nothing is quoted.

Inspired by
Comment in Ask HN: Who is using OpenClaw?
Creator
devashish86 on Hacker News
Published on
Hacker News
Original date
Apr 16, 2026
License
Not openly licensed (the poster keeps all rights), so we quote nothing and link to the original
Originally for
OpenClaw
AI dry run
Oct 23, 2026: passed. Dry-run tested by our AI test agent on sample input. PASS: Gates held on desk check; no live sends, bookings or purchases.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails.
Verified
Oct 23, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.

Are you the creator? Claim or correct this page.

More in Operations & Admin

All 203 Operations & Admin playbooks →

Related playbooks

Operations & AdminWorkflow✓ AI dry run

Collect office snack requests in Slack and stage a cart

Turn teammates' grocery requests in a chat channel into a clean list, get it approved, and have an assistant build the delivery cart and post the link back.

Credit: nekuda

Under an hour
Operations & AdminPrompt✓ AI dry run

Give your agents one shared wiki that only one agent edits

Replace scattered agent memories with a single Markdown wiki in git: every agent reads a short daily brief from it, agents and scripts file sourced notes into an inbox instead of editing pages, one steward agent checks and merges those notes each morning, and private agent memory keeps only personal preferences.

Credit: Siim Tolk

An afternoon+
Operations & AdminWorkflow✓ AI dry run

Scope an email triage agent after a month's trial

Get fast triage and routine drafts across several inboxes while keeping negotiations, sensitive relationships and CC choices with you.

Credit: Jarvis Specter

An afternoon+
Operations & AdminRoutine✓ AI dry run

Assemble the Friday report from your source files

Have the recurring Friday report assembled from your source files into the real template, needing only a light edit.

Credit: Suraj Khaitan

Minutes