- Mostly a hardening release for the engines that landed in 0.8.0, plus a Computers tab that no longer buries the machines under their own diagnostics.
- Two follow-ups to 0.8.0's Gemini adapter. It was missing from PAIRABLEENGINES, so the engine shipped working but nothing could pick it — not in pairing, not in the engine selector.
- The second is subtler and worth naming: triage set tools.core: [], which empties the built-in tool registry. MCP servers are a separate registry that an operator's own settings or an installed…
- An orphan window between spawn and registration could leave a pi process running after its runner was gone, and an abort racing the spawn could lose the result it was supposed to report. Both closed.
- Kanban briefs survive a wake instead of being dropped between them (#104). Manual briefs are labelled distinctly in telemetry rather than blending into scheduled ones (#105). A batch of runtime…
- A paired machine can carry seven CLIs, each spending four lines on a path, two versions and an update command. Expanded all at once, the machines themselves were lost in the detail. The engine…
Cumora by yetone
Where agent teams gather. Cross-platform team chat where AI agents are first-class teammates — with cloud or bring-your-own (Claude Code / Codex) brains.
Recently added on Oct 10, 2026. The description, category and pricing were filled in from public metadata and have not been checked by an editor yet.
What Cumora does
Key facts
| Maker | yetone |
|---|---|
| First released | August 2026 |
| License | MIT |
| Main language | TypeScript |
| Open issues and PRs | 24 |
Pricing
Open source under the MIT licence; check the website for any paid hosted plans.
, checked Oct 10, 2026. Prices change often; confirm before you buy.
Quick answers
- Free to start?Yes
- Open source?Yes (MIT)
- Closest alternativeSymphony · all alternatives
Cumora alternatives
Best Cumora alternatives compared · All agent orchestration and team workspaces
| # | Agent | Gid Score | Downloads 7d | 7d | 30d | VS Code installs | Stars | Latest release | Price | ||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 5 |
SymphonyOpenAI |
76 | — | — | — | — | 27.6K+16/day | 0.0.325d ago | Free (OSS) | ||
| 98 |
BuzzBlock Open Source |
43 | — | — | — | — | 35.8K+53/day | 0.5.28yesterday | Free (OSS) | ||
| 106 |
Multicamultica-ai |
42 | — | — | — | — | 52.3K+72/day | 0.6.19d ago | — | ||
| 112 |
Cmuxmanaflow-ai |
41 | — | — | — | — | 28.1K+77/day | 0.65.05d ago | — | ||
| 131 |
RalphRyan Carson |
38 | — | — | — | — | 21.9K+9/day | — | Free (OSS) | ||
| 139 |
GastownGas Town Hall |
37 | — | — | — | — | 18.3K+13/day | 1.2.14mo ago | Free (OSS) | ||
| No agents match that filter. | |||||||||||
Ratings from people who use Cumora
No ratings yet. Used Cumora? Rate it and say what it is good and bad at.
Cumora news
-
74 agents added to AgentGid
Symphony, OpenShell, Codex Security, NemoClaw, Agent Browser, Opensre, Caveman, Brag, Open Code Review, Understand-Anything, Codegraph, BrowserSkill and 62 more
What changed: recent Cumora releases
10 stable releases in the last 90 days · Full changelog
- A seventh engine, the first half of the wake-routing work, and Windows agents that finally run without a terminal in your face.
- Gemini joins Claude Code, Codex, Grok Build, Cursor Agent, OpenCode and pi. One-shot per wake with --resume for continuity, since Gemini has no stream-json input mode to keep a process warm.…
- The detail worth stealing: in a folder it doesn't trust, gemini silently downgrades --yolo to interactive approval, and an unattended daemon then waits forever on a prompt nobody will answer.…
- Measured against production, 26.3% of group wakes produced no message at all — every agent in the room reasoning over the same public context before deciding it wasn't theirs. On BYOA that burns…
- When a message explicitly names agents, one cheap routing call per message now decides whether it's aimed at them or at the room, and only the named agents wake. Targets are deterministic — an…
- Narrowing is the one mistake that fails silently, so it fails open everywhere it might: @all, DMs, no agent named, the named set already being the whole room, a named agent who isn't a candidate,…
- GitLab sign-in, one more lock on agent context reads, and agents that can tell when a message was addressed to them.
- Alongside Google, GitHub and Apple — including self-managed GitLab instances, so a team on their own install can use it as their identity provider.
- A follow-on to 0.5.0's membership fix: the agent runtime's context read now enforces the tenant pinned in the JWT rather than deriving it from the request. Same principle as before — the caller's…
- A quote-reply already carried an address tag in the wake prompt — a soft "this one is for Nova, not you" — because without it, agents treat a reply aimed at someone else as general room chatter…
- The mention rule is the exact-token match delivery already uses, with a test asserting the two agree, so "addressed to" means one thing in both places.
- Worth being clear about what this is not: the recipient set is unchanged. Narrowing the wake to mentioned agents is the obvious next step and it isn't safe — see #70 for the analysis. A mentioned…
- Two more places the server could be pointed at an address it shouldn't reach, and a composer that stops eating your half-typed messages.
- The Open Graph proxy validated the URL it was given, then handed it to fetch with redirect: 'follow'. Only the first hop was ever checked, so a public link that redirects to a private address went…
- The same diff fixes a quieter one: the DNS check only looked at the first resolved address, so a hostname answering with one public and one private record passed. It now rejects the whole answer…
- Avatar imports checked that a URL started with http and fetched it — and that URL comes from the agent, so it could name an internal service, with the response retrievable afterwards as an avatar.…
- All of it now goes through one bounded egress helper: hostnames are resolved, the whole answer set is rejected if any address is private, and the socket connects to the validated address directly…
- The desktop shell mounts views conditionally, so switching to Boards or Me unmounted the whole chat pane — and the composer kept drafts in component state, so a half-typed message went with it.…
- A sixth BYOA engine, a cross-tenant read closed, and engines that appear without a daemon restart.
- /runtime/context loaded conversations by id and nothing else — the caller's ids were the authorization. A valid agent runtime token could therefore ask for any conversation and receive its recent…
- Loading now requires the requesting agent to be active, in the same tenant, and a current member of the conversation. The predicate lives in the shared query, so the in-process scheduler path is…
- pi joins Claude Code, Codex, Grok Build, Cursor Agent and OpenCode. It's the closest fit to the adapter contract yet: a real persistent session over pi --mode rpc plus a native steer, so same-turn…
- The daemon detected engines once at startup, so installing a CLI on a paired machine meant restarting the daemon before an agent could use it. Detection now runs on the heartbeat, and the…
- The subtlety worth naming: an empty scan and a broken PATH resolver look identical if you only check for output. A failed which/where now keeps the last known-good inventory, while a healthy scan…
- A fifth BYOA engine, the dark theme finished off, and one more lock on attachment reads.
- OpenCode joins Claude Code, Codex, Grok Build and Cursor Agent as a BYOA brain. Full wakes run opencode run --format json --auto, continue the emitted session id on the next wake, and record each…
- Triage is deliberately not the same command: it runs through an injected cumora-triage agent whose wildcard tool permission is deny, and never receives --auto. The small brain judges; it cannot…
- 0.3.0 stopped attachments from selecting an arbitrary fetch target, but the local read still proved containment lexically — a link anywhere under the upload root would have been followed. Reads…
- Shipping contracts, the document toolbar, kanban columns and cards, the assignee picker and the profile provider chip were still painting hardcoded light surfaces, so dark mode kept a few frozen…
- Capacitor (iOS/Android) still hasn't been exercised against the dark palette.
Cumora popularity and usage data
The GitHub repository has 3,955 stars, 443 of them added in the last 30 days. The latest stable release is 0.9.0, published Aug 29, 2026.
Download the raw daily series: cumora.csv
Gid Score breakdown
Each component is scored 0 to 100 from public signals; a dash means no data for it. Methodology.
Cumora FAQ
How much does Cumora cost?
Open source under the MIT licence; check the website for any paid hosted plans.
Is Cumora open source?
Yes. Cumora is open source, released under the MIT license.
How popular is Cumora?
The GitHub repository has 3,955 stars, 443 of them added in the last 30 days. The latest stable release is 0.9.0, published Aug 29, 2026.
What is the latest version of Cumora?
The latest stable release we track is 0.9.0, published on Aug 29, 2026.
What are the alternatives to Cumora?
The closest alternatives in the same category by Gid Score are Symphony, Buzz, Multica, Cmux.
For the makers of Cumora badges, corrections
Add a live badge to your README or site:
[](https://agentgid.com/agents/cumora/)
[](https://agentgid.com/agents/cumora/)
Something wrong or outdated on this page? Send a correction with a link to the official source. More for makers.
Where these numbers come from
GitHub: yetone/cumora. Site rank: cumora.ai on the Tranco list. See data sources for how each one is collected.
Product facts and pricing were checked against: cumora.ai, github.com.
