$git clone https://github.com/cuga-project/cuga-agentBuilding a domain-specific enterprise agent from scratch is complex and requires significant effort: agent and tool orchestration, planning logic, safety and alignment policies, evaluation for performance/cost tradeoffs and ongoing improvements. CUGA is a state-of-the-art general
| 1 | <picture> |
| 2 | <source media="(prefers-color-scheme: dark)" srcset="/docs/images/cuga-dark.png"> |
| 3 | <source media="(prefers-color-scheme: light)" srcset="/docs/images/cuga-light.png"> |
| 4 | <img alt="CUGA" src="/docs/images/cuga-dark.png"> |
| 5 | </picture> |
| 6 | |
| 7 | <div align="center"> |
| 8 | |
| 9 | # CUGA: Configurable Generalist Agent — Agent Harness for the Enterprise |
| 10 | |
| 11 | ### Start with a generalist. Customize for your domain. Deploy faster! |
| 12 | |
| 13 | Building a domain-specific enterprise agent from scratch is complex and requires significant effort: agent and tool orchestration, planning logic, safety and alignment policies, evaluation for performance/cost tradeoffs and ongoing improvements. CUGA is a state-of-the-art generalist agent designed with enterprise needs in mind, so you can focus on configuring your domain tools, policies and workflow. |
| 14 | |
| 15 | --- |
| 16 | |
| 17 | [](https://huggingface.co/spaces/ibm-research/cuga-agent) |
| 18 | |
| 19 | [](https://www.python.org/) |
| 20 | [](https://docs.cuga.dev/docs/sdk/cuga_agent/) |
| 21 | []() |
| 22 | [](https://docs.cuga.dev) |
| 23 | [](https://discord.gg/aH6rAEEW) |
| 24 | |
| 25 | [%20on-AppWorld-gold?style=for-the-badge)](https://appworld.dev/leaderboard) |
| 26 | [%20on-WebArena-gold?style=for-the-badge)](https://docs.google.com/spreadsheets/d/1M801lEpBbKSNwP-vDBkC_pF7LdyGU1f_ufZb_NWNBZQ/edit?gid=0#gid=0) |
| 27 | |
| 28 | </div> |
| 29 | |
| 30 | --- |
| 31 | |
| 32 | > **Why CUGA?** — A generalist agent harness for the enterprise: wire your APIs and MCP servers, tune reasoning and task modes, and govern behavior with policies—without rebuilding orchestration from scratch. |
| 33 | > |
| 34 | > | Feature | How | |
| 35 | > |---------|-----| |
| 36 | > | **MCP, OpenAPI & LangChain tools** | [`mcp_servers.yaml`](src/cuga/backend/tools_env/registry/config/mcp_servers.yaml) · `CugaAgent(tools=[...])` | |
| 37 | > | **Reasoning modes** (fast / balanced / accurate) | `[features] cuga_mode` in [`settings.toml`](src/cuga/settings.toml) · [`configurations/modes/`](src/cuga/configurations/modes/) | |
| 38 | > | **Hybrid API + browser tasks** | `[advanced_features] mode = 'hybrid'` · Playwright + [browser extension](src/frontend_workspaces/extension/readme.md) | |
| 39 | > | **Multi-agent (CugaSupervisor)** | `cuga start demo_supervisor` · `[supervisor]` in [`settings.toml`](src/cuga/settings.toml) | |
| 40 | > | **A2A & remote agents** | External agent entries in supervisor config · [CugaSupervisor](https://docs.cuga.dev/docs/sdk/cuga_supervisor) | |
| 41 | > | **Policies & HITL** | [Policies SDK](https://docs.cuga.dev/docs/sdk/policies/) — Intent Guard, Playbook, Tool Approval, Tool Guide, Output Formatter | |
| 42 | > | **Manage & publish** | `cuga start manager` · draft tools, MCP, LLM, and policies in the web UI, then **publish** a versioned config for production chat ([details](#manage-publish-and-self-hosting)) | |
| 43 | > | **Reflection** | `[advanced_features] reflection_enabled` in [`settings.toml`](src/cuga/settings.toml) | |
| 44 | > | **Langflow** | Low-code visual workflows — integrates with CUGA ([langflow.org](https://www.langflow.org/)) | |
| 45 | > | **Knowledge** (RAG) | `enable_knowledge=True` (default) · ingest PDFs/Office/HTML/Markdown via **Docling** · **agent-level** + **session-level** scopes · `cuga start demo_knowledge` · [details](#knowledge-base) | |
| 46 | > | **Agent skills** | `SKILL.md` under `.cuga/skills` (default) · **`cuga start demo_skills`** (`sandbox_mode = "native"` by default, or **`opensa |