AgentSeek Desktop

AgentSeek Desktop is the desktop UI for discovering AgentSeek templates, creating isolated instances, managing .env files, running lifecycle commands, and inspecting logs.

Requirements

• Node.js 24 or newer.
• Rust stable and the Tauri system prerequisites.
• Python 3.12+ and uv for agentseek CLI installation.
• agentseek executable on PATH (or installed via pipx install agentseek).
• Docker and Docker Compose V2 only when Docker deployment is selected.

Development

git clone https://github.com/agentseek-ai/agentseekd.git
cd agentseekd
npm install
npm run tauri dev

Use the browser preview when working only on the UI:

npm run dev

The preview uses local browser storage and simulated AgentSeek operations. Packaged Tauri builds use real CLI and filesystem commands.

Build

npm run build
cargo check --manifest-path src-tauri/Cargo.toml
npm run tauri build

During repository development, the backend invokes the local checkout with uv run --project <repo> agentseek. Installed builds prefer an agentseek executable and otherwise fall back to uvx agentseek.

Set AGENTSEEK_CLI to an executable path to override this resolution.

Testing

Frontend Tests

npm run test          # run once
npm run test:watch    # watch mode
npm run test:coverage # with coverage report

Covers the Tauri API client, i18n helpers, and shared types.

Rust Tests

cd src-tauri
cargo test

124 tests covering CLI resolution, env bridge, instance management, lifecycle patches (apt mirror, PyPI mirror, GitHub mirror, CORS, async shim, OpenVINO model conversion), and template rendering.

Template-based tests require cached cookiecutter templates at ~/.cookiecutters/agentseek/templates/. If absent, these tests are skipped gracefully.

E2E Template Tests

End-to-end tests that create a real agentseek instance for each template, configure .env, install dependencies, start services, and verify conversation via the API.

# Run all E2E tests (auto-starts Mock API server if no API key)
npm run test:e2e

# Skip Docker-dependent templates
npm run test:e2e:skip-docker

# CI mode (skip templates needing local hardware)
npm run test:e2e:ci

The built-in Mock API server (tests/e2e/mock-api-server.py) provides an OpenAI-compatible API with zero dependencies and zero cost. When no E2E_API_KEY is set, the script automatically starts the mock server.

To test with a real LLM, set these environment variables:

export E2E_API_KEY="sk-..."
export E2E_API_BASE="https://api.openai.com/v1"
export E2E_MODEL="openai:gpt-4o-mini"
...
npm run test:e2e

Supported Templates

Template	Protocol	Docker	Notes
bub/default	Bub AG-UI	No
deepagents/default	Bub AG-UI	No
deepagents/research	LangGraph	No	Needs TAVILY_API_KEY (mock provided)
deepagents/sandbox	LangGraph	No	Needs DAYTONA_API_KEY (optional)
deepagents/content-builder	LangGraph	No
langchain/default	Bub AG-UI	Yes	Phoenix + SeekDB
langchain/cli-remote	LangGraph	No
langchain/agentic-rag	LangGraph	Yes
langchain/agentic-rag-hybrid	LangGraph	Yes
langchain/agentic-rag-openvino	LangGraph	Yes	Downloads models via task models
langchain/markdown-messages	LangGraph	No