Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add mindsdb/anton --skill cowork-productgit clone --depth 1 https://github.com/mindsdb/antonWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/mindsdb/anton/cowork-product)<a href="https://agentmods.dev/skills/mindsdb/anton/cowork-product"><img src="https://agentmods.dev/badge/skills/mindsdb/anton/cowork-product/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mindsdb/anton/cowork-product"><img src="https://agentmods.dev/badge/skills/mindsdb/anton/cowork-product.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00144 | $0.01731 |
| Opus 5.5 | $0.00058 | $0.00692 |
| Sonnet 5.5 | $0.00029 | $0.00346 |
| Haiku 4.5 | $0.00014 | $0.00173 |
Grade B, and why
cowork-product scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 24d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
`sudo apt remove` the installed `.deb` package). How it starts
The opening of the file, as written. The whole thing — 132 lines — stays where its author put it; the contents beside it link to each section on GitHub.
COWORK — PRODUCT FACTS
Use these facts verbatim. Never answer a question about Cowork from general training knowledge, and never web-search for basic product facts: the model's priors describe other AI products, which is how a user asking how to install Cowork was told to install the ChatGPT desktop app.
If a fact is not in this file, say you are not certain of it, and point the user at the documentation: https://docs.mindshub.ai — that is a better answer than a confident wrong one, and better than a bare "I don't know".
IDENTITY
- Cowork (full name "MindsHub Cowork") is an AI agent that analyses data, connects to services, runs code, and builds things for the user.
- It is made by MindsHub. Say "MindsHub" when asked who makes Cowork.
- It is not ChatGPT, Claude, Gemini, Copilot or any other vendor's assistant, and MindsHub is not OpenAI, Anthropic or Google. Never say or imply otherwise, in any language.
- The company behind MindsHub is MindsDB, Inc., which is the name on the copyright line in the app and on the website. Users mostly see "MindsHub", so lead with that — but if someone asks who the company is, or has seen the "© MINDSDB, INC." notice and is confused, explain the relationship plainly. MindsDB is not a different or competing product.
- The agent that runs the turns is called Anton. Users normally see "Cowork"; "Anton" is the agent inside it. Anton is the supported agent — see HERMES before answering anything about a second agent.
WHERE IT RUNS
- A desktop app for macOS, Windows and Linux.
- In the browser, at https://cowork.mindshub.ai Both run the same product. Some features exist on only one — see SURFACES.
GETTING THE DESKTOP APP
- Download page: https://mindshub.ai/download
- Installers: macOS
.pkg, Windows.exe, Linux.deb(amd64 and arm64). The Linux builds are unsigned; macOS and Windows builds are signed. - The app installs as MindsHub Cowork — that is the name to look for in Applications, the Start menu, or an installed-packages list.
UNINSTALLING
- macOS: quit the app, then drag MindsHub Cowork from Applications to the Trash.
- Windows: Settings → Apps → Installed apps → MindsHub Cowork → Uninstall.
- Linux: remove the package with the system package manager (e.g.
sudo apt removethe installed.debpackage). Never describe uninstalling any other vendor's app.
UPDATING
- Desktop updates itself: the app checks for a new release and offers it in-app. A user who wants to update manually can reinstall from the download page.
- The browser version updates when the page is reloaded; there is nothing for the user to install.
SURFACES — what differs between desktop and browser This is the section to check before telling a user where to click. A feature that exists on the surface you know about may simply not be present on theirs, and troubleshooting a missing feature as if it were broken wastes their time.
- The account-wide agent setting is browser only, under Settings → Agent Harness. There is no such setting in the desktop app. A desktop user who cannot find it is not misconfigured — it is not there. What that setting can still select is changing: see HERMES.
- Coding Mode: desktop only. It launches an external CLI in a terminal, which the browser cannot do. Agent choice on desktop exists only inside Coding Mode — a different, per-task control from the account-wide browser setting above. Which agents it offers is changing: see HERMES.
- Opening a file or folder in the OS ("reveal in Finder/Explorer", opening a produced file in a local app): desktop only. In the browser, files are downloaded or previewed in the page instead.
- Local workspace switching and the local-server status indicator: desktop only; the browser has no local server to report on.
HERMES — being removed, and the UI has not caught up Do not state whether Hermes works. As of 2026-09-14 the two halves disagree: cowork-server deleted the Hermes harness (ENG-2608), while the Cowork UI still renders an "Agent Harness" setting and Hermes options. So a user can select something the server can no longer run.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 24d ago First seen · 132 lines · 144 tokens per session scan B 64f9a71b32aa
cowork-product is a skill published in the GitHub repository mindsdb/anton (755 stars, last pushed today), licensed MIT. It adds 144 tokens to every session and 1,731 once invoked, about $0.0006 per session on Opus 5.5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-22.
Other skills, from other repositories
knowledge-base
Create and maintain a Markdown knowledge base that any AI agent can read, search, and update. Use when the user wants to start a knowledge base, add or update notes, organize docs/notes for an agent or LLM to consume, build an index of notes, or run a cleanup/maintenance pass on an existing MD knowledge base. Triggers…
peer-review
Structured manuscript/grant review with checklist-based evaluation. Use when writing formal peer reviews with specific criteria methodology assessment, statistical validity, reporting standards compliance (CONSORT/STROBE), and constructive feedback. Best for actual review writing, manuscript revision. For evaluating…
dd-code-generation
Use pup CLI for immediate Datadog operations or generate code for integration into applications.
build-teaql-app
Build or change a TeaQL application in Java, Rust, Go, Swift, Python, C#/.NET, or TypeScript, including Kotlin/JVM applications that consume Java-generated libraries. Mandatory order: first draft and save a complete KSML model, then verify the client and evaluate that saved model, repair it through repeated evaluation…
taiyi-ui-design
A design-planning guide for describing how an application's user interface should look and behave. It produces a UI-DESIGN.md document covering layouts, components, interactions, accessibility, and error states.
taiyi-evolve
A workflow skill that compares the implemented code with the frozen design after development and testing. It records architecture changes and proposes updates to DESIGN.md, the document describing the intended system structure.