CLI usage reference

nasiko builds, tests, deploys, and monitors agents against a control plane. See CLI installation to install it and the quickstart for the core workflow. State lives in ~/.nasiko/config.json (clusters, active cluster, tokens) and ~/.nasiko/.env (local-stack config).

Every command operates against the active cluster, selected with nasiko use <name>. Run nasiko <command> --help for full flags.

Setup

Command	Purpose
`nasiko up`	Start a full local platform (infra + control plane)
`nasiko down`	Stop the local platform
`nasiko connect <url> [--name <name>]`	Register a control plane by URL
`nasiko use <name>`	Switch the active cluster
`nasiko clusters`	List registered control planes
`nasiko status`	Active cluster’s health and metrics
`nasiko auth login` / `status` / `logout` / `whoami`	Authentication

nasiko connect https://nasiko.example.com --name prod
nasiko use prod
nasiko auth login

Create

Command	Purpose
`nasiko new [template] [name]`	Scaffold a new agent project (interactive if template is omitted)
`nasiko skill add/remove/list/search/info`	Manage agent skills (tools) attached to a project
`nasiko card [description] [--dir <dir>]`	Generate or update `AgentCard.json`
`nasiko validate [dir]`	Check `AgentCard.json`, `Dockerfile`, and `src/` are in order

nasiko new                          # interactive: pick a framework
nasiko new openai my-agent          # skip the prompts
nasiko validate ./my-agent

Test

Command	Purpose
`nasiko build [dir] [--tag <tag>] [--platform <platform>]`	Build the agent’s Docker image
`nasiko run [dir] [--port <port>]`	Build and run the agent locally (default port `8000`)
`nasiko chat [target] [message] [--agent <name>] [--tui] [--session-id <id>] [--resume <id>]`	Send an A2A message
`nasiko sessions` / `create-session` / `history` / `delete-session`	Manage chat sessions

nasiko run .                                   # build + run on :8000
nasiko chat http://localhost:8000 "hello"      # talk to the local instance directly
nasiko chat --tui                              # full-screen chat TUI

nasiko chat’s target can be a URL, an agent name or UUID, or orchestrator. Omit it to talk to the routing engine. A lone argument containing whitespace is treated as the message, not the target.

Operate

Command	Purpose
`nasiko push <image> [--name <name>]`	Build + push an image to the cluster registry (no deploy)
`nasiko deploy <image-or-dir> [--name <name>] [--port <port>] [--env-file <file>] [-e KEY=VALUE]`	Build + push + deploy
`nasiko upload [source] [--name <name>] [--version <v>] [--port <port>]`	Upload source; the server builds and deploys it
`nasiko ps`	List running agents
`nasiko logs <agent> [-n <count>] [-f]`	Stream agent logs (`-f` live-tails via SSE)
`nasiko stop/start/restart <agent>`	Lifecycle control (`restart` picks up new secrets/env)
`nasiko scale <agent> <replicas>`	Scale to N replicas
`nasiko rm --name <agent>`	Terminate and deregister (accepts a raw UUID positionally instead of `--name`)
`nasiko secrets set/get/ls/rm [--agent <name>]`	Encrypted secrets (vault-wide, or per agent)
`nasiko observe sessions/session/trace-detail/span/...`	Traces, tokens, and cost

nasiko deploy . --port 8000 -e LOG_LEVEL=debug
nasiko logs my-agent -f
nasiko scale my-agent 3
nasiko secrets set OPENAI_API_KEY sk-... --agent my-agent
nasiko observe session <session-id>

Pass an image reference instead of a directory to nasiko deploy to skip the build. nasiko push builds and pushes without starting the container — it does still register a registered-status catalog entry for the agent (visible in nasiko ps/agents ls), it just never runs.

Multi-agent

Command	Purpose
`nasiko maf workflow create --name <name> --step <task> [--agent <name\|uuid\|->] ...`	Define a workflow (repeat `--step`/`--agent` per step; `-` auto-assigns)
`nasiko maf workflow run <id-or-name> [--wait]`	Queue an execution
`nasiko maf workflow get <id>`	Read one workflow
`nasiko maf execution list` / `get <id>` / `result <id>`	Inspect executions

nasiko maf workflow create --name "research-and-summarize" \
  --step "Research recent news on topic X" --agent research-agent \
  --step "Summarize the research into 3 bullet points" --agent -   # "-" auto-assigns
nasiko maf workflow run research-and-summarize --wait
nasiko maf execution result <execution-id> --json

Full flag reference: nasiko maf --help. See MAF for the workflow model and REST API.

LLM configuration

Command	Purpose
`nasiko llm-config providers/create/attach/get`	Per-agent named LLM configs (provider, model, fallback, tuning)
`nasiko model-registry ls/set`	Platform-wide strength-level → model table (requires superuser)

nasiko llm-config create --name support-cheap --provider openai --model gpt-4o-mini
nasiko llm-config attach support-bot support-cheap

Full command set: nasiko llm-config --help. See routing engine for how these fit into agent selection.

Integrations

Command	Purpose
`nasiko agents ls/get/deploy/search/info/frameworks/list-uploaded/chat`	Agent registry + lifecycle operations
`nasiko github status/repos/connect/disconnect/clone`	GitHub OAuth connection + clone-to-deploy
`nasiko registry connect/disconnect/status/search/list`	Connect to and browse the artifact registry
`nasiko mcp catalog/connect/connections/disconnect/connector/credential/oauth/agent-tools`	MCP Gateway connector and tool-permission management

nasiko github clone myorg/my-agent-repo
nasiko registry search "invoice processing"
nasiko mcp catalog
nasiko mcp connect --toolkit <name>       # from the catalog output

What you’re connecting to

Entity	What it is
Control plane	Deploys agents, routes queries, proxies A2A traffic, serves the observability and registry APIs
Embedded OCI registry	The control plane’s `/v2/*` endpoint — stores your deployed agent images
Artifact registry	A separate service for templates and skills, used by `nasiko new`, `skill search`, and `registry`

The CLI only operates on agents you own or have been granted. See access control.

Getting Started

CLI

ADLC

Dashboard

Platform

MCP Gateway

MCP Hub

Agents in Artifact Registry

Setup

Create

Test

Operate

Multi-agent

LLM configuration

Integrations

What you’re connecting to

​Setup

​Create

​Test

​Operate

​Multi-agent

​LLM configuration

​Integrations

​What you’re connecting to

Setup

Create

Test

Operate

Multi-agent

LLM configuration

Integrations

What you’re connecting to