Own your tag.
No vendor lock-in.
Open source Claude Tag alternative.
An AI worker in your Slack and 
Lark, all agents, all models.
1/3Connect your workspace
Pick where the agent lives. One bot covers the whole team.
LarkSlack
2/3Choose your model
Runs on the plan you already pay for.
Claude Code
Codex
Kimi
DeepSeek
Z.ai+Any provider
3/3You’re set up.
It runs on your machine. Now put it in the room.Invite it to # product-eng
Product Eng
11
MessagesDocsFilesAnnouncements…
Serena invited OpenTag to the group
Priya4:09 PM
@OpenTag last week’s signups?
OpenTag BOT4:09 PM
Pulled from Google Analytics
1,240 signups
Docs site 62% · was 21%
Sam4:11 PM
@OpenTag signup button is dead on mobile.
OpenTag BOT4:11 PM
Opened an issue on GitHub
#482 Signup button dead on mobile
opened · 2 linked
🙏 12🚀 1
Hold fn to talk
Message Product Eng
Product Eng
11
MessagesDocsFilesAnnouncements…
Sam9:41 AM
@OpenTag I just joined — where is the project at?
OpenTag BOT9:41 AM
Pieced together from 14 threads, 3 channelsLaunch is Monday, pricing page still open
Kept by OpenTag · on your own machine
🙏 12👀 1
Hold fn to talk
Message Product Eng
Connect your stack
The same tag reaches all of them, from any channel.
GitHub
Slack
Notion
Gmail
Google Calendar
Google Drive
Google Sheets
Linear
Jira
GitLab
Figma
HubSpot
Mailchimp
Airtable
Zoom
Stripe
Postgres
Google Analytics
Connect your stack
The same tag reaches all of them, from any channel.
GitHub
Slack
Notion
Gmail
Google Calendar
Google Drive
Google Sheets
Linear
Jira
GitLab
Figma
HubSpot
Mailchimp
Airtable
Zoom
Stripe
Postgres
Google Analytics
Used by people at
OpenAI
Google
GUAZI.COM
Google
GUAZI.COM
Open source, so you own every layer.
Every part of it is yours and nothing is hidden: the tag, the memory, the plan, all portable, none of it locked to us.
Own your memory.
No context lock-in.
Bring your own subscription.
No model lock-in.
You already pay for this. Pay for it once.
Bring the token provider you already pay for. The software is free and we take no cut.
…
Memory that stays put.
One copy of what your team knows, held across every channel and shared by the people and the agents in them. Decisions and designs are written down once and reused, instead of explained again to whoever asks next.
It is git-native: markdown files in Git, each with an owner. Agents propose updates after a task, and you review them the way you review code.
Connect your stack.
The connectors are your agent’s own. Whatever it can already reach, the room can reach.
It reads and it acts — pull up the issue, check the calendar, open the pull request — in the thread, with your own credentials on your own machine.
What builders say.
“It finally feels like a build tool that works the way I want.”
JoshPlatform Engineer, Adobe
“Running six agents in parallel and reviewing every diff in one place. This is the workflow I wanted.”
BrandonSoftware Engineer, Reddit
“As a designer, the UI is the first thing I judge. This one actually holds up.”
Angela FUI/UX Designer, Figma
“It's like adding five engineers to a two-person startup, overnight.”
Bruce ZSerial Entrepreneur
“Agents triage my repo's issues and open PRs while I sleep. For a solo maintainer, that's huge.”
Sam HOpen-Source Maintainer
“Beautiful UI, and it handles all the git worktree stuff for you.”
SammySoftware Engineer, Meta
The same job, without the middleman.
Scroll to compare
| Capability | OpenTag | Claude Tag | openclaw | Viktor |
|---|---|---|---|---|
| Open source | Apache 2.0 | |||
| Team feature | One agent the whole channel uses | |||
| Model support | Any model, through your own key | Anthropic models only | Bring your own models | Its own agent's models |
| Bring your own subscription | Any API key, Claude & Codex plans | DMs only | hosted, subscription or local | — |
| Local first | Mac, Linux | Mac, Windows, Linux | ||
| Option to self-host | Docker | npm or from source | ||
| Pricing model | Free · you pay your provider | Usage balance | Free, self-hosted | Workspace credits |
Questions.
What does it cost?
The software is free and Apache 2.0 licensed. You pay your model provider directly at their rates, plus whatever your own hosting costs. No credits and no markup on inference.
Where does our memory actually live?
In a directory on the machine you run OpenTag on, as plain files you can open, edit, and delete. It is never uploaded to us, because there is no “us” in the path — no OpenTag-operated server sees your messages or your memory.
Which models does it work with?
Whichever your agent supports — Anthropic, OpenAI, Kimi, DeepSeek, local models. We don't tie you to one provider, so you can switch without redoing your setup.
Which messengers does it support?
Lark and Slack today. More messengers are on the roadmap and will land in the open repo before they land on this page.
Can I really use my Claude or ChatGPT subscription?
OpenTag supports signing in with a Claude or Codex plan as well as with an API key. Worth knowing: consumer subscriptions are sold for use in the provider's own apps, and whether third-party access is permitted is governed by each provider's terms, which change. API keys are the unambiguous path and what we recommend for a shared workspace. Read your plan's terms and decide for your own team.
Do I need to be technical to set it up?
Whoever sets it up runs the agent on a machine they control and creates a Lark or Slack bot. After that, using it is just @-ing a bot in a thread — the rest of the team sets up nothing.
Does it do org-level permissions and admin?
Not yet. If your team needs them, open an issue — that's how we prioritise.
Just tag it.
The future of work is humans and agents building side by side.