Menu
Use when the user wants to install cuga, start the demo/web UI, or asks what `cuga start` does or which service/mode to run.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Use when the user wants to govern agent behavior with a cuga policy - block/redirect an intent, add step-by-step playbook guidance, require approval before a tool runs, enhance a tool's description, or reshape agent output format.
Use when the user wants to write Python code that creates or invokes a CugaAgent or CugaSupervisor, e.g. "build an agent with cuga", "how do I call agent.invoke", "set up a multi-agent supervisor".
Use when the user wants a running CugaAgent to load a new capability at runtime via cuga's own skill system, e.g. "add a skill to my agent", "create a .cuga/skills SKILL.md", "how does load_skill work".
Use when the user wants to install, launch, or build something with the cuga agent framework (pip package `cuga`), or asks "how do I use cuga" / "how do I build an agent with cuga" — routes to the right cuga-harness-kit skill instead of guessing.
Use when the user wants a cuga agent to ingest, search, or answer questions from documents (PDF/DOCX/XLSX/PPTX/HTML/Markdown/images) - RAG / knowledge base features.
Use when the user wants to give a cuga agent a new capability via a Python function, REST/OpenAPI service, or MCP server, e.g. "add a tool", "connect this API to cuga", "register an MCP server".
Based on SOC occupation classification
| name | cuga-install-and-launch |
| description | Use when the user wants to install cuga, start the demo/web UI, or asks what `cuga start` does or which service/mode to run. |
For a new app project, initialize with uv and add cuga as a dependency so it is recorded in pyproject.toml:
uv init my-cuga-app
cd my-cuga-app
uv add cuga
For an existing project that already has a pyproject.toml, run uv add cuga from the project root. If the user only wants a quick install inside an already-active virtualenv, uv pip install cuga is acceptable.
(cuga requires Python >= 3.10, < 3.14.) For working inside a checkout of the cuga-agent repo itself instead of the published package:
git clone https://github.com/cuga-project/cuga-agent.git
cd cuga-agent
uv venv --python=3.12 && source .venv/bin/activate
uv sync
Set LLM API keys before starting anything. Use a project-local .env; see docs/cuga-env-api-keys.md for provider-specific AGENT_SETTING_CONFIG, MODEL_NAME, and API-key examples.
uv run cuga start <service>
Valid <service> values: demo, demo_skills, demo_crm, demo_docs, demo_health, demo_knowledge, demo_supervisor, travel_agent, manager, registry, appworld.
demo / demo_crm / demo_supervisor / demo_knowledge / demo_docs — launch the chat web UI at https://localhost:7860 plus a tool registry service on port 8001, each preset wiring up different sample tools (CRM+email, multi-agent supervisor, RAG knowledge, docs) so you can try the matching capability immediately.demo_skills — same UI, with cuga's own runtime skill-loading enabled (see cuga-build-cuga-skill).manager — a draft/publish UI: edit agent config (tools, MCP servers, LLM, policies) as a draft, try it, then publish a versioned config for production chat.registry — just the tool registry service (OpenAPI/MCP config), no chat UI.appworld / travel_agent — specific benchmark/example scenarios.Other useful commands:
cuga stop <service> — stop a running service.cuga status — show what's currently running.cuga doctor — environment/dependency sanity check.cuga viz — trajectory viewer dashboard (see cuga-debug-trajectory).There is no cuga policy or cuga knowledge CLI subcommand — policies and knowledge are managed through the Python SDK (agent.policies.*, agent.knowledge.*, see cuga-author-policy / cuga-knowledge-rag) or through the manager web UI.
Once the UI is up, point the user at cuga-build-agent for writing their own Python code against the SDK, or have them just chat with the demo agent directly in the browser.