Reported News
kwant: self-hosted multi-bot AI team in one Rust binary (0.7.4)
kwant's README describes "a team of bots that share chats, hand work to each other with an @mention and get it done on their own, on a schedule or while you're away," shipped as one Rust binary with an embedded React app under AGPL-3.0. The project site says 0.7.4 released Nov 7, with 29 releases since Oct 25, and the GitHub repo had 217 stars when we checked on Nov 7, 2026.
- Changes reported
- 6
- listed below, each from a source we read
- We tested
- Not tested yet
- reported from sources; we did not run it ourselves
- Gaps we flag
- 3
- things we could not verify
- Sources
- 3
- 1 primary, linked below
What changed
| # | What the source says | Status |
|---|---|---|
| 01 | New open-source product: kwant is a self-hosted multi-bot AI desk created on GitHub on Oct 22, 2026. The README says bots share chats, hand work with an @mention, and run on a schedule or while you are away, with separate spaces for work and home. | Reported |
| 02 | One binary: the README says it is written in Rust (Tokio, Axum, SQLite) with a React web app embedded in the binary. Downloads cover macOS (universal Apple silicon and Intel) and Linux x86_64 (.deb or AppImage), plus Docker Compose for a server; no Windows app yet. | Reported |
| 03 | Verified outcomes: work appears as Mission board cards, and a task turns Verified only when kwant runs the check itself, never on a bot's word. Risky steps such as git push or network calls show an approval card; the README says payments always ask. | Reported |
| 04 | Accounts: it runs Claude, Codex or Grok CLI logins you already have, plus API keys or your own compatible server, with failover when an account runs out of usage. | Reported |
| 05 | Coding-agent MCP (0.7.4): Claude Code, Codex, Cursor and other MCP clients can hand bots work with a per-connection token that reaches only the bots you pick and cannot approve anything. | Reported |
| 06 | Release cadence and license: the homepage states 0.7.4 released Nov 7 and 29 releases since Oct 25 under AGPL-3.0. When we checked on Nov 7, 2026 the GitHub repo had 217 stars and 14 forks; 0.7.4 also adds a daily spending stop (tokens or dollars) and merge-request reviews on GitLab, Gitea and Forgejo. | Reported |
reported by the sourcemeasured or run by usnot verified
Primary source: kwant repository and README (GitHub). We are the fast, independent index; the vendor’s own page is the source of record for this release.
In the source’s words
“kwant gives you a team of bots that share chats, hand work to each other with an @mention and get it done on their own, on a schedule or while you're away.”
“It's written in Rust (Tokio, Axum, SQLite) with a React web app embedded in the binary.”
“a task turns Verified only when kwant runs the check itself, never on a bot's word.”
Early verdict
Worth a look if you want a local or VPS multi-bot desk with handoffs, approvals and verified outcomes instead of a hosted chat product; note the AGPL-3.0 license and that we have not run it.
Basis: Based only on the GitHub README, homepage and 0.7.4 release notes as read on Nov 7, 2026.
What we have not verified
- We did not install or run kwant, open a space, or exercise approvals, routines or Dev mode.
- We did not measure binary size, idle RAM, Docker image sizes or the Laya model router.
- We did not verify commercial-license terms beyond the README's AGPL-3.0 and commercial-license mention.
This brief is Reported: written from the sources below, not from running the release ourselves.
Sources
- kwant repository and README (GitHub)primary source · fetched 2026-11-07
- kwant homepage (kwant.si)source · fetched 2026-11-07
- kwant 0.7.4 release (GitHub)source · fetched 2026-11-07
Related
Spotted an error? See our corrections policy or email the address on the contact page. Markdown version for agents: /news/kwant-self-hosted-ai-agent-team.md
Want these briefs in your Google Top Stories? Add Agentic Weekly as a preferred source on Google.
Newsletter
The Weekly Verdict. Free, every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.