by openyak
Open-source local-first AI agent for desktop work. No account, no telemetry: use local models with Ollama/Rapid-MLX or bring your own provider key.
# Add to your Claude Code skills
git clone https://github.com/openyak/openyakLast scanned: 5/5/2026
{
"issues": [
{
"type": "npm-audit",
"message": "axios: Axios has a NO_PROXY Hostname Normalization Bypass that Leads to SSRF",
"severity": "high"
},
{
"type": "npm-audit",
"message": "follow-redirects: follow-redirects leaks Custom Authentication Headers to Cross-Domain Redirect Targets",
"severity": "medium"
},
{
"type": "npm-audit",
"message": "lodash: lodash vulnerable to Code Injection via `_.template` imports key names",
"severity": "high"
}
],
"status": "WARNING",
"scannedAt": "2026-05-05T06:19:42.234Z",
"semgrepRan": false,
"npmAuditRan": true,
"pipAuditRan": true
}openyak is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by openyak. Open-source local-first AI agent for desktop work. No account, no telemetry: use local models with Ollama/Rapid-MLX or bring your own provider key. It has 703 GitHub stars.
openyak returned warnings in SkillsLLM's automated security scan. It has no critical vulnerabilities, but review the flagged issues in the Security Report section before adding it to your workflow.
Clone the repository with "git clone https://github.com/openyak/openyak" and add it to your Claude Code skills directory (see the Installation section above).
openyak is primarily written in Python. It is open-source under openyak on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh openyak against similar tools.
No comments yet. Be the first to share your thoughts!
Requires a passing catalog security scan. Resolve the flagged issues and resubmit to enable featuring.
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
OpenYak is built for people who want an AI agent that lives on their machine instead of behind another hosted workspace account.
New in the v1.5 release candidate. Native Computer currently ships for macOS and Windows. The rest of OpenYak continues to support Linux, but Linux Computer Use is not a v1.5 release gate.
OpenYak gives the Agent two purpose-built, shared workspaces instead of treating every interface as a stream of blind coordinate clicks:
Choose Use → Computer to pin the task to a native app, Browser to pin it to the managed Browser, or Auto to let the Agent choose the appropriate tools for that turn. The selected mode stays visible in the composer, and every Computer or Browser tool observation identifies the active surface before you open its live workspace. For example:
Review the launch checklist, switch to Notes, then verify the release page in the Browser.
Keep me in the live workspace and hand control back whenever I take over.
The demo uses sample apps and websites whose approvals are already granted so the short walkthrough can show the actual task sequence: user message, streaming Agent response, Computer and Browser tool observations, opening each live workspace, takeover and return, then the Agent's final verification. In normal use, macOS asks for Accessibility and Screen Recording access; Windows keeps the target app visible on the active desktop, and UAC secure desktop is outside the runtime. If you select a cloud model, the relevant app or page state needed for the task is sent directly to that provider as model context.
Read the Computer Use architecture, supported actions, safety model, and platform status.
| Ask OpenYak to... | It should give you... |
|---|---|
| Read a long memo | Executive brief, risks, owners, next actions, and a send-ready email |
| Analyze a workbook | Budget vs. actual variance, drivers, anomalies, and finance talking points |
| Review a deck | Slide-by-slide story, evidence gaps, speaker notes, and decision ask |
| Synthesize several files | One board brief that reconciles memo, budget, deck, and PDF context |
| Work in a native app or website | A shared live workspace where the Agent acts and you can take over at any time |
| Split work across agents | Parallel child-agent tasks with links back to each focused session |
| Continue the same thread | RACI, 30-day plan, agenda, and follow-up drafts without restating context |
| Recover from an error | Clear next step when upload, auth, or file parsing fails |
OpenYak can turn a dense memo into a structured brief that is ready for a manager, team update, or follow-up email.
Use spreadsheets as working inputs, not screenshots. Ask for budget variance, forecast risks, owner-level action items, and meeting-ready talking points.
OpenYak can synthesize several files in the same thread and open a right-side artifact panel for reusable briefs, plans, diagrams, and structured outputs.
Turn on Ultra for complex work and describe the goal normally. The parent Agent decides when to launch 2–4 focused AgentRuns in durable child Sessions, follows their live state, exchanges input when needed, and synthesizes their persisted results in the original task. There is no task-batch form to configure.
The parent-scoped Subagents view keeps Active and Done work inside the same task shell. Open a child transcript, inspect its status and output, then return to the parent while Progress, Outputs, Subagents, Sources, Inputs, and Context stay tied to the same task.
Real office work rarely fits in one message. OpenYak is designed for follow-ups, revisions, and long conversations where the important context needs to remain available.
| Platform | Architecture | Formats |
|---|---|---|
| macOS | Apple Silicon / Intel | .dmg, .app |
| Windows | x64 | .exe installer |
| Linux | x64 | .deb, .rpm |
Download the latest release or visit open-yak.com/download.
To try Computer Use now, download the v1.5 release candidate.
Linux users can also read LINUX.md for requirements and troubleshooting.