Unified CLI for the AgentKit ecosystem
Manage all your AI agent services from one tool.
npx @agentkitai/agentkit-cli initThis creates an agentkit.config.yaml in your project, sets up docker-compose.yml, and scaffolds your workspace.
Interactive project setup. Asks for project name, language (TypeScript/Python), and which services to enable.
$ agentkit init
? Project name: my-agent
? Language: typescript
? Enable AgentLens? yes
? Enable Lore? yes
...
✅ Project initialized!
Show the status of all configured services.
$ agentkit status
Service | Status | Port | Version
---------------|------------|-------|--------
agentlens | ✅ running | 3000 | 1.2.0
lore | ✅ running | 8765 | 1.4.2
agenteval | ⚪ disabled | - | -
Options:
-c, --config <path>— Config file path (auto-detected if omitted)-t, --timeout <ms>— Connection timeout (default: 3000)-w, --watch [seconds]— Live-refresh the table every N seconds (default 2; Ctrl+C to exit)
up starts the governed stack — it is not a thin docker compose up -d
wrapper. On first run it generates fresh secrets
(LORE_API_KEY / ADMIN_API_KEY / JWT_SECRET) via ensureSecrets(), defaults
to the governance profile when none is given, and (with --wait) seeds demo
data on first run once the stack is healthy. down and logs stay thin wrappers
around docker compose in your stack directory:
$ agentkit up # fresh secrets + governance profile, then compose up -d
$ agentkit up --profile minimal # override the default profile
$ agentkit up --wait # block until healthy, then first-run demo seed
$ agentkit logs -f lore # docker compose logs -f lore
$ agentkit down -v # docker compose down -v (removes volumes)
Options:
up:-p, --profile <name>(minimal | governance | full, default: governance),--no-detach,-w, --wait(block until the stack is healthy),-t, --timeout <ms>(--waittimeout, default: 120000)down:-v, --volumes(also remove volumes)logs [service]:-f, --follow- all:
-c, --config <path>(locates the stack'sdocker-compose.yml)
Run diagnostic checks on your setup.
$ agentkit doctor
✅ Config file
✅ Docker daemon
✅ agentlens container
❌ lore health → Check logs with `docker compose logs lore`
Options:
-c, --config <path>— Config file path (auto-detected if omitted)-t, --timeout <ms>— Connection timeout (default: 3000)
The agentkit.config.yaml file:
projectName: my-agent
language: typescript
services:
agentlens:
enabled: true
port: 3000
lore:
enabled: true
port: 8765
agentgate:
enabled: false
formbridge:
enabled: false
agenteval:
enabled: false| Field | Type | Description |
|---|---|---|
projectName |
string | Project name (required) |
language |
"typescript" | "python" |
Project language |
services.<name>.enabled |
boolean | Whether the service is active |
services.<name>.port |
number | Override default port |
services.<name>.version |
string | Pin a specific version |
| Service | Default Port | Description |
|---|---|---|
| AgentLens | 3000 | Observability and monitoring for AI agents |
| Lore | 8765 | Knowledge management and RAG pipeline |
| AgentGate | 3002 | API gateway and rate limiting |
| FormBridge | 3003 | Form generation and data collection |
| AgentEval | — | Evaluation and benchmarking (CLI-only) |
When you run status or doctor without --config, the CLI walks up the directory tree (up to 10 levels) looking for agentkit.config.yaml. If none is found, it prints:
No agentkit.config.yaml found. Run `agentkit init` to get started.
Contributions are welcome! Fork the repo, make your changes, and open a pull request. For major changes, open an issue first to discuss what you'd like to change.
| Project | Description | |
|---|---|---|
| AgentLens | Observability & audit trail for AI agents | |
| Lore | Cross-agent memory and lesson sharing | |
| AgentGate | Human-in-the-loop approval gateway | |
| FormBridge | Agent-human mixed-mode forms | |
| AgentEval | Testing & evaluation framework | |
| agentkit-cli | Unified CLI orchestrator | ⬅️ you are here |