Eigent Open-Source Cowork Desktop
Local-first open-source AI cowork desktop for multi-agent collaboration and automation, as a free alternative to Claude Cowork and Codex.
- Source repo
- eigent-ai/eigent
- Stars
- ★ 16k
- Last updated
- 2d ago
- License
- Apache-2.0
- Primary language
- TypeScript
- FA score
- 25/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platform
- Cost
- Free tier plus a paid hosted plan
- Setup effort
- Medium · a few setup steps
- You'll need
- Typical use
- A developer automates monthly dev reports by running local models (DeepSeek via Ollama) to review GitHub PRs and generate Word summaries and Slack updates.
- Not a fit if
- Teams that need a fully offline setup with no account registration
- Users unwilling to run a Python backend and local model inference
- Production teams needing mature docs and stable enterprise support
- Source review
- 25/100 · Major gaps 8 safety controls not found
What does this agent do, and when should you use it?
Eigent is an open-source desktop application built on the CAMEL-AI framework, enabling orchestration of multi-agent workflows. It emphasizes local deployment, giving users full control over data without a cloud account. The app includes built-in browser, terminal, and MCP toolkits, and is model-agnostic, supporting Ollama, vLLM, LM Studio, and cloud APIs. Users work with single or multiple agents on research, writing, debugging, and can schedule automated workflows. Eigent is 100% open source from day one, with transparent development and community contribution.
Eigent provides a desktop app and local backend server that coordinates work via the CAMEL multi-agent framework. Users create tasks; agents execute multi-step workflows in parallel, using browser, terminal, and MCP tools. It integrates local inference models (Ollama, vLLM) or cloud APIs, keeping data local. The frontend is React/TypeScript on Electron; the backend is FastAPI with OAuth 2.0. Automation features allow scheduled triggers like monthly report generation.
- A developer automates monthly dev reports by running local models (DeepSeek via Ollama) to review GitHub PRs and generate Word summaries and Slack updates.
- A user asks Eigent to organize a cluttered desktop by inspecting files and sorting them into logical folders.
- A game developer coordinates parallel agents to build 10 HTML5 games with scoring and difficulty progression.
- An ML team audits CI failures with multi-agent orchestration that fetches logs, compares golden values, and produces structured reports.
- An enterprise integrates a ticket management system, importing local data and generating statistical charts.
- A solo developer builds a 3D platformer with Eigent coordinating coding and testing.
How do you install or deploy this agent?
Clone the repo, run npm install and npm run dev for cloud-connected quick start. For local deployment, follow the server/README_EN.md guide to configure the backend and local models (Ollama, vLLM).
How do you use this agent?
Launch the app, register/login (cloud mode) or configure local backend. Create tasks, choose single or multi-agent workflows, and connect to models. Use built-in browser and terminal tools, or schedule tasks for automation.
What are this agent's strengths and limitations?
- 100% open source with auditable code and transparent development.
- Local-first deployment ensures data privacy and no cloud dependency.
- Model-agnostic architecture supports Ollama, vLLM, LM Studio, and cloud APIs, avoiding vendor lock-in.
- MCP and skill integration for extensibility.
- Built-in browser and terminal toolkits ready to use.
- Local deployment requires Python environment and model inference setup, which may be complex.
- Cloud quick-start requires account registration and network connectivity.
- Multi-agent quality depends on model performance and may need tuning.
- Enterprise features like SSO are sales-led and may not be fully open.
- Early-stage project with limited documentation and community resources.
How does this agent compare with similar options?
Eigent is positioned as an open-source replacement for Claude Cowork and Codex, emphasizing local deployment and model freedom, though detailed feature comparisons are not in the README.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Eigent Open-Source Cowork Desktop This agent | 25 · Major gaps | Desktop appFreemium | ★ 16k | 2d ago | TypeScript | — |
| Skales | 36 · Major gaps | Desktop appFreemium | ★ 2.2k | 6d ago | — | ChatGPT · OpenAI API · Claude API |
| WorkBuddy Practical Guide | 27 · Major gaps | Web appFree | ★ 3.5k | 1mo ago | TypeScript | — |
| AgentChat | 29 · Major gaps | Self-hosted serviceFree + model costs | ★ 921 | 1d ago | Python | — |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows SECURITY.md and Apache-2.0 license, but no specific implementation of least privilege, user confirmation, data flow transparency, sensitive data handling, dependency security, external effects, rollback, or source attribution. All trust criteria scored 0 due to lack of evidence.
Self-consistency: README claims local deployment and zero external dependencies, but package.json includes many dependencies and build workflows require external services (e.g., AWS S3), contradiction, score 1. Dependency availability: dependency list is complete but no lock files or integrity checks, score 1. Failure messages: no error handling or user-friendly failure messages found, score 0.
Audience and scenarios: README describes multiple use cases (developers, enterprise, cloud), score 2. Capability boundaries: no explicit limits of agent capabilities, score 1. Trigger precision: no trigger mechanism described, score 0. Environment fit: supports multiple platforms (macOS, Windows, Linux) but no system requirements, score 1.
Information architecture: README has clear structure with TOC and sections, score 2. Install notes: provides local deployment and quick start steps, score 2. Naming stability: version exists but no naming convention, score 1. Examples and FAQ: multiple use case links but no FAQ, score 2. Known limitations: not mentioned, score 0. License: Apache-2.0, score 2. Versioning and changelog: version number but no changelog, score 1. Maintenance responsibility: contribution guide and community channels but no explicit maintainers, score 1.
Output usability: no description of output format or integration, score 1. Marginal value: as open-source alternative with unique multi-agent features, score 2. Cost benefit: no performance or cost data, score 1.
Claim traceability: README claims not linked to specific code or tests, score 1. Cross-source corroboration: no external verification, score 0. Fact-inference separation: some claims (e.g., 'zero setup') lack evidence, score 1.
- Not found in source: least-privilege scopingGrant only what the task needs: a dedicated account or read-only token, scoped to specific directories and repos.
- Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
- Not found in source: data-flow disclosureWatch which external services it contacts (proxy or firewall logs) and keep sensitive data out until you know where it goes.
- Not found in source: sensitive-data handlingUse dedicated, low-privilege, revocable API keys — never production credentials — and keep secrets out of logs.
- Not found in source: dependency securityPin versions and run a dependency audit (npm audit, pip-audit) before installing; prefer running it in a container.
- Not found in source: disclosed external effectsEstablish which external systems it writes to, sends to or changes, and verify with test accounts or repos before production.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- Not found in source: verifiable attributionInstall from the official repo or registry and check the publisher and URL to avoid look-alike packages.
- No least privilege or user confirmation mechanisms found; agents may have excessive permissions.
- Dependency security not verified; recommend checking dependency locking and vulnerability scanning.
- README's 'zero external dependencies' contradicts the many dependencies in package.json.
- No rollback mechanism provided; operations may be irreversible.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
Release notes excerpt
Eigent v1.0.6 introduces redesigned Automations and model settings, adds a work-role step to onboarding, and makes Sessions more reliable when switching views, reloading, or resuming interrupted work. Highlights - Redesigned Automations — Start from six example workflows, see upcoming runs, and inspect results in a dedicated run dashboard. Pause, resume, or request a run from the automation list. Schedule editing now provides clearer local-time previews and validation, while preserving existing schedules the editor cannot represent. Scheduled runs require Eigent to remain open on thi …View on GitHub ↗ -
Release notes excerpt
Eigent v1.0.5 improves Session recovery, task queues, and the tools you use to inspect ongoing work. It also adds Meta Model API support, makes Space resources easier to configure, and introduces a limited, opt-in preview of parallel Session execution. Highlights - More reliable Session recovery — Recover task state, conversation history, elapsed time, and completed output more consistently after interrupted connections or app restarts. Resume Sessions with their accepted model configuration and handle model switching after usage limits more reliably. …View on GitHub ↗ -
Release notes excerpt
Eigent v1.0.4 Eigent v1.0.4 introduces a redesigned way to manage Skills and Connectors, while strengthening reliability across multi-turn agent work, workspace checkpoints, and app shutdown. The new management experience makes resources easier to browse, inspect, and configure, while a focused set of fixes helps Tasks complete more predictably. Highlights - Redesigned Skills and Connectors - Browse and manage Skills in a library-style dashboard with dedicated detail views. - Inspect source and access details, enable or disable Skills, and review their files from one place. …View on GitHub ↗