Documentation

View .mdOptimized for LLMs — paste directly into ChatGPT, Claude, or Cursor.

CLI reference

The agntz CLI ships inside @agntz/sdk. It creates YAML manifests, runs agents locally, publishes local state, and manages hosted runs, traces, and evals from the terminal.

# Run without installing
npx @agntz/sdk --help

# Or install globally
npm i -g @agntz/sdk
agntz --help

For the first local workflow, start with CLI getting started.

Command map

Command	Local?	Hosted?	Auth?	Purpose
`create`	-	✓	No	Generate YAML from a description through the hosted builder.
`validate <path>`	✓	-	No	Validate YAML, duplicate ids, and cross-file agent refs.
`run <path>`	✓	-	No	Run a local YAML file or single-agent directory.
`run <id>`	-	✓	Yes	Run a hosted agent by id.
`login` / `logout` / `whoami`	-	✓	Mixed	Manage hosted API credentials.
`publish`	✓	✓	Yes	Import local agents, sessions, and memrez memory into hosted storage.
`runs`	-	✓	Yes	List, inspect, stream, or cancel hosted runs.
`traces`	-	✓	Yes	List, inspect, or delete hosted traces.
`eval`	-	✓	Yes	Run hosted evals and inspect eval runs or latest scores.

Every command supports terminal help:

agntz create --help
agntz validate --help
agntz run --help
agntz login --help
agntz publish --help
agntz runs --help
agntz traces --help
agntz eval --help

Auth and configuration

Hosted commands read credentials in this order:

AGNTZ_API_KEY
~/.agntz/config.json, written by agntz login

API URL resolution uses:

command --url where supported
AGNTZ_API_URL
saved config
https://api.agntz.co

Local runs do not require an agntz API key. They use provider keys from your process environment, such as OPENAI_API_KEY, ANTHROPIC_API_KEY, or other keys required by the manifest's model/tool configuration.

`create`

agntz create "<description>" [options]

Generates a YAML manifest by calling the hosted agent-builder. No login is required.

Flag	Description
`-o, --output <path>`	Write the manifest to a specific path. Default: `./agents/<id>.yaml`.
`--stdout`	Print YAML to stdout instead of writing a file.
`--current-manifest <path>`	Revise an existing manifest instead of starting fresh.
`--url <apiUrl>`	Override the builder API URL for this call.
`-h, --help`	Show command help.

Examples:

agntz create "Answer support questions in a concise tone" -o ./agents/support.yaml

agntz create "Add an HTTP order lookup tool" \
  --current-manifest ./agents/support.yaml \
  -o ./agents/support.yaml

agntz create "Classify inbound leads by urgency" --stdout > ./agents/lead-classifier.yaml

create validates that the builder returned YAML, parses the manifest to get its id, creates parent directories as needed, and prints the local run command to try next.

`validate`

agntz validate [path] [--json]

Validates one YAML file or recursively scans a directory. The default path is ./agents; dependency, build-output, coverage, and hidden directories are ignored during recursion. Directory validation also rejects duplicate agent ids and unresolved pipeline, spawnable, or agent-as-tool references. The command exits with status 1 when any file fails or no agent manifests are found.

--json prints a complete machine-readable report, including per-file errors, warnings, and aggregate counts:

agntz validate ./agents
agntz validate ./agents/support.yaml --json

For editor completion, associate YAML files with https://agntz.co/schemas/agent-manifest.schema.json or use the packaged schema at @agntz/core/schema.

`run`

agntz run <path-or-id> [options] [input...]

Runs an agent. The target determines local vs hosted mode unless you force a mode.

Target shape	Mode
`./agents/support.yaml`	Local YAML file
`agents/support.yml`	Local YAML file
`./agents`	Local directory, only if it contains exactly one manifest
`support`	Hosted agent id

Flag	Description
`--input <text>`	Input string. Use `--input -` to read stdin.
`--session <id>`	Reuse a session id across calls.
`--stream`	Stream reply/complete/error events instead of buffering the final output.
`--local`	Force local execution.
`--remote`	Force hosted execution.
`-h, --help`	Show command help.

Input resolution:

--input value > trailing positional text > piped stdin > empty string

Examples:

# Local file
agntz run ./agents/support.yaml --input "How do I reset my password?"

# Local file with stdin
cat ticket.txt | agntz run ./agents/support.yaml

# Local file with persistent conversation state
agntz run ./agents/support.yaml --session user-42 --input "My email changed"

# Hosted agent id
agntz run support --input "Hello" --remote

# Stream hosted or local output
agntz run ./agents/support.yaml --input "Walk me through this" --stream

Local runtime boundary: agntz run ./agents/support.yaml constructs a local SDK client with agntz({ agents: "<manifest-dir>" }). It can run agents whose requirements are satisfied by YAML plus environment configuration. If the agent declares local tools or resource providers that need application code, call @agntz/sdk from your service and pass tools / resources there.

`login`, `logout`, and `whoami`

agntz login --key <apiKey> [--url <apiUrl>]
agntz logout
agntz whoami

login writes credentials to ~/.agntz/config.json with owner-only permissions. logout removes that file. whoami prints the resolved API URL and a masked key source.

Examples:

agntz login --key ar_live_...
agntz login --key ar_live_... --url https://agntz-worker.example.com
AGNTZ_API_KEY=ar_live_... agntz whoami
agntz logout

Browser-based login is not implemented in the current CLI. Paste an API key from the hosted or self-hosted dashboard.

`publish`

Migrates local manifests, persisted sessions, and memrez memory into hosted storage. Requires AGNTZ_API_KEY or agntz login.

agntz publish [all|agents|sessions|memory...] [options]

Common options:

Flag	Description
`--agents <dir>`	Local agents directory. Default: `./agents`.
`--db <path>`	Local SDK SQLite store for sessions.
`--memory-db <path>`	Local memrez SQLite store. Default: `./memory.db` or `./memrez.db` if present.
`--dry-run`	Report what would be imported without writing hosted state.
`--include-superseded`	Include superseded memory entries.

Examples:

agntz publish all --dry-run
agntz publish agents sessions memory --db ./agntz.db --memory-db ./memory.db
agntz publish memory --memory-db ./memrez.db

`runs`

Hosted run management. Requires AGNTZ_API_KEY or agntz login. Output is JSON.

agntz runs list   [--agent <id>] [--status <s>] [--limit <n>] [--cursor <c>]
agntz runs get    <runId>
agntz runs stream <runId> [--since <seq>]
agntz runs cancel <runId>

Examples:

agntz runs list --agent support --limit 20
agntz runs get run_123
agntz runs stream run_123 --since 10
agntz runs cancel run_123

runs stream emits the multiplexed event stream for a hosted run subtree. --since <seq> resumes from a sequence number.

`traces`

Hosted trace management. Requires AGNTZ_API_KEY or agntz login.

agntz traces list   [--agent <id>] [--status <s>] [--limit <n>] [--cursor <c>]
agntz traces get    <traceId>
agntz traces delete <traceId>

Examples:

agntz traces list --agent support --status failed --limit 10
agntz traces get trace_123
agntz traces delete trace_123

`eval`

Hosted eval management. Requires AGNTZ_API_KEY or agntz login.

agntz eval run    <evalId> [--dataset <id>] [--version <agentVersion>]
agntz eval runs   [--agent <id>] [--eval <id>] [--dataset <id>] [--status <s>] [--limit <n>] [--cursor <c>]
agntz eval cancel <runId>
agntz eval scores [--agent <id>] [--eval <id>] [--dataset <id>] [--version <createdAt>]
agntz eval get    <evalId>

Examples:

agntz eval run support-quality --dataset refund-cases
agntz eval runs --agent support --limit 10
agntz eval scores --eval support-quality --dataset refund-cases
agntz eval get support-quality

Current CLI boundary

The CLI covers manifest generation, recursive validation, local execution, hosted execution, state publishing, hosted run/trace inspection, and hosted eval execution. It does not provide project scaffolding, an interactive playground, or a Studio launcher. Use the SDK for application runtime wiring and the hosted app for managed agent editing.

Exit behavior

Exit code	Meaning
`0`	Success
`1`	Argument, auth, network, builder, validation, or runtime error

Most commands write human-readable errors to stderr. agntz validate --json and agntz publish --json provide structured CLI output; use @agntz/sdk for deeper local integration or @agntz/client for hosted execution.

← Previous

@agntz/client

Hosted cloud

CLI reference

Command map

Auth and configuration

create

validate

run

login, logout, and whoami

publish

runs

traces

eval

Current CLI boundary

Exit behavior

`create`

`validate`

`run`

`login`, `logout`, and `whoami`

`publish`

`runs`

`traces`

`eval`