Eigent Open Source Cowork Release Notes: Skills, Models & a More Powerful Agent Platform 🚀

This sprint pushes Eigent further toward being a fully extensible, multi-model, multi-agent platform. From a full Skills System to expanded model support and smoother UI polish, these updates make Eigent more customizable, more powerful, and more enjoyable to use. This release also strengthens the cowork experience and reinforces Eigent’s long-term direction around open source cowork.

As always, these improvements are community-driven. Let’s dive in.

---

🧩 A Full Skills System + New Agents Tab

This is a foundational upgrade.

Huge thanks to @4pmtong, @Pakchoioioi, and @Douglasymlai for shipping one of the biggest architectural improvements in this cycle.

Eigent now includes a complete Skills System, deeply integrated across backend toolkits, Electron file management, and the frontend UI. Alongside it comes a brand new Agents tab in the main navigation.

What’s new:

• Upload, delete, and manage skills directly from the UI
• Scope control for skills across agents
• Built-in example skills to get started immediately:
  • code-reviewer
  • data-analyzer
  • report-writer
• SkillToolkit integration across all worker agents:
  • Browser
  • Developer
  • Document
  • Multi-Modal
  • and more
• A new Agents tab centralizing:
  • Models
  • Skills
  • Memory

This makes agents significantly more modular and configurable, a major step toward a true agent platform.

🔗 PR: https://github.com/eigent-ai/eigent/pull/1221

---

🎥 Built-in Audio & Video Playback in File Browser

Small feature. Big usability win.

Shoutout to @Zephyroam for adding native multimedia playback directly into Eigent’s file explorer.

Eigent now supports playing audio and video files directly in the Folder view, without switching apps or leaving your workflow.

Because of this PR, you now get:

• Native audio playback inside the file browser
• Native video playback inside the file browser
• A clean, properly styled media preview layout

This is especially useful for agents generating media outputs or when reviewing multimedia artifacts as part of a workflow.

🔗 PR: https://github.com/eigent-ai/eigent/pull/1286

---

🧠 Multi-Modal Attachments Across All Workers

Huge thanks to @BitToby for making cross-agent collaboration even smoother.

Multi-modal attachments (like images) now work consistently across:

• Browser
• Developer
• Document

This means image context survives worker handoffs, enabling much richer collaboration patterns between agents.

You now get:

• Reliable image attachments across worker flows
• Better shared context between agents
• More seamless multi-modal reasoning

It’s a big step toward making agent cooperation feel natural instead of fragile—and toward making open source cowork with AI agents feel seamless.

🔗 PR: https://github.com/eigent-ai/eigent/pull/1196

---

🎨 Redesigned Worker Node UI

Polish matters, especially when you’re coworking with agents all day.

Big thanks to @Douglasymlai for redesigning the worker node component with smoother interactions and better usability.

Improvements include:

• Refined animations and motion
• Clean micro-interactions and transitions
• One-click copy button for task reports
• Fixed broken design tokens from a recent merge

The result? A worker node that feels more responsive, more intentional, and more professional.

🔗 PR: https://github.com/eigent-ai/eigent/pull/1140

---

⚙️ Smarter Model Configuration Handling

Another thoughtful improvement from @Douglasymlai.

Model configuration now handles edge cases more gracefully and detects expired setups automatically.

What’s improved:

• Proper handling of model title display edge cases
• Automatic detection of expired model configurations
• Clear warnings when attention is required

This prevents silent failures and gives users better visibility into their model setup.

🔗 PR: https://github.com/eigent-ai/eigent/pull/1201

---

🤖 Gemini 3.1 Pro Preview Now Available

Model flexibility keeps expanding.

Thanks to @fengju0213, Eigent now supports Gemini 3.1 Pro Preview as a selectable model option.

In addition to adding the model, this PR also cleans up redundant locale strings across all supported languages.

Because of this PR, you now get:

• Gemini 3.1 Pro Preview in the model selector
• Cleaner, deduplicated i18n settings

More model choice. Less config friction.

🔗 PR: https://github.com/eigent-ai/eigent/pull/1314

---

🌐 Expanded Model Platform Support

Eigent continues to become model-platform agnostic.

Huge thanks to @Wendong-Fan for adding support for three major LLM platforms:

• Grok
• Mistral
• SambaNova

You now get:

• Grok platform integration
• Mistral platform integration
• SambaNova platform integration
• Full backend + frontend integration with tests

This significantly broadens the ecosystem of models that work out of the box with Eigent.

🔗 PR: https://github.com/eigent-ai/eigent/pull/1321

---

❤️ Community Keeps Raising the Bar

This sprint delivered:

• A full Skills System
• Expanded multi-model support
• Better multi-modal handoffs
• Polished worker UI
• Smarter model configuration handling
• Native multimedia playback

Every single one of these improvements came from contributors who cared enough to improve what they use.

That’s the power of open source—and the power of open source cowork.

If you’ve been thinking about contributing skills, models, UX, agents, tooling, or cowork workflows, this is your moment.

Let’s keep building.