CLI reference

This page is generated from the current Morphz CLI schema and matches the current binary.

Command index

Command	Description
morphz exec	Run one prompt and print the final reply
morphz resume	Reattach an existing or recently active Session
morphz serve	Start the HTTP/WebSocket runtime and embedded Dashboard
morphz dashboard	Start the Dashboard and open it in the default browser
morphz edge	Pair and run an outbound Execution Node
morphz edge pairing-code	Create a short-lived pairing code for the current Principal
morphz edge nodes	List Execution Nodes owned by the current Principal
morphz edge revoke	Revoke one paired Execution Node
morphz edge local-leases	List Provider-local capability leases for this Node
morphz edge revoke-local-lease	Revoke one Provider-local capability lease
morphz edge pair	Pair this device with a Morphz Gateway
morphz edge run	Run the authenticated outbound Edge worker
morphz edge rotate-key	Rotate this Node's device identity key
morphz edge status	Show the paired Node credential and local Target identity
morphz target	Inspect and administer Execution Targets
morphz target list	List Targets visible to the current Principal
morphz target show	Inspect one Target
morphz target enable	Enable one Target
morphz target disable	Disable one Target
morphz target authorize	Restrict a Target to an Agent, Context or Thread scope
morphz target authorizations	List scoped Target authorizations
morphz target revoke-authorization	Revoke one scoped Target authorization
morphz lease	Inspect and revoke Target capability leases
morphz lease list	List active capability leases
morphz lease revoke	Revoke one capability lease
morphz execution	Inspect and control durable physical Execution Jobs
morphz execution list	List Execution Jobs
morphz execution show	Inspect one Execution Job
morphz execution output	Read durable stdout/stderr chunks for one Job
morphz execution cancel	Request cancellation of one Job
morphz setup	Open guided model provider setup
morphz provider	Inspect and verify model providers
morphz provider list	List catalog and configured providers
morphz provider test	Verify a provider catalog, stream and tool call
morphz provider show	Show one effective Provider Instance
morphz provider set	Validate and persist a Provider Instance TOML file
morphz provider account	Manage provider authentication accounts
morphz provider account list	List account configuration and runtime state
morphz provider account login	Start an OAuth login
morphz provider account complete	Complete or poll an OAuth login
morphz provider account logout	Revoke a stored OAuth login
morphz provider account set	Validate and persist non-secret Auth Account TOML
morphz provider account enable	Enable one account
morphz provider account disable	Disable one account
morphz provider account test	Diagnose one account through a compatible Model Route
morphz model	Discover or select models
morphz model list	List models exposed by a provider
morphz model use	Persist the default provider and model
morphz model refresh	Refresh and verify the remote catalog for one Model Route
morphz model route	Manage logical Model Routes
morphz model route list	List effective Model Routes
morphz model route show	Show one Model Route
morphz model route set	Validate and persist a Model Route TOML file
morphz model route test	Diagnose route resolution, account auth and provider health
morphz profile	Inspect or select configuration profiles
morphz profile list	List available profiles
morphz profile show	Show the resolved contents of a profile
morphz profile use	Select the default profile
morphz context	Inspect persistent Cognitive Contexts
morphz context list	List Cognitive Contexts
morphz context show	Show one Cognitive Context
morphz context status	Show Context state, Sessions and active work
morphz context audit	Verify the Context Mind projection by replaying Events
morphz context recall-index	Inspect or rebuild the derived lexical Recall index
morphz context recall-index inspect	Show Recall index capability and document counts
morphz context recall-index rebuild	Rebuild the derived Recall index from Events and Mind
morphz context recall	Search Context memory or traverse one Mind Frame lineage
morphz context recall search	Search indexed Event and Mind Frame documents
morphz context recall frame	Traverse Mind Frame sources and relations
morphz scheduler	Inspect authoritative Scheduler state
morphz scheduler show	Show Threads, activations, jobs, approvals and schedules
morphz scheduler thread	Inspect and control one durable Thread
morphz scheduler thread show	Show one Thread causal chain and structured Outcome
morphz scheduler thread pause	Pause a Thread
morphz scheduler thread resume	Resume a Thread
morphz scheduler thread cancel	Cancel a Thread
morphz scheduler thread supersede	Cancel the current generation and continue with a corrected intent
morphz session	Manage Sessions within a Context
morphz session list	List Sessions
morphz session show	Show one Session
morphz session create	Create a Session within a selected Context
morphz session resume	Reattach an existing or recently active Session
morphz agent	Manage persistent Agents
morphz agent list	List Agents
morphz agent show	Show one Agent
morphz agent create	Create an Agent with a Root Context and initial Session
morphz harness	Install and inspect versioned Harness packages
morphz harness list	List installed Harness versions
morphz harness show	Show one exact installed Harness version
morphz harness install	Validate and install a .hns file or directory
morphz objective	Manage long-lived Objectives
morphz objective list	List Objectives in a Context
morphz objective show	Show one Objective
morphz objective create	Create and run a long-lived Objective
morphz objective edit	Replace an Objective goal using revision fencing
morphz objective pause	Pause an Objective
morphz objective resume	Resume an Objective
morphz objective cancel	Cancel an Objective
morphz trajectory	Export and verify portable Agent Trajectory Bundles
morphz trajectory export	Export authoritative Runtime facts as an Agent Trajectory Bundle
morphz trajectory verify	Validate an untrusted Agent Trajectory Bundle
morphz trajectory episode	Derive a permission-checked training Episode
morphz storage	Inspect and migrate Runtime storage authority
morphz storage migrate-cognitive-store	Explicitly synchronize cognitive state into the selected Store
morphz experiment	Inspect and verify explicitly gated experimental features
morphz experiment list	List compiled and enabled experimental features
morphz experiment check	Require one experiment to be compiled and enabled
morphz job	Inspect or cancel delegated Sub Agent jobs
morphz job list	List delegated jobs
morphz job cancel	Cancel a delegated job and its descendants
morphz config	Inspect resolved configuration and provenance
morphz config show	Print the resolved configuration
morphz config check	Validate all loaded configuration layers
morphz config path	List loaded configuration files in precedence order
morphz config explain	Explain the source of every resolved value
morphz update	Update Morphz from a verified GitHub Release
morphz update status	Check the latest release without changing any files
morphz update rollback	Restore the version retained by the last update
morphz doctor	Check storage, workspace, permissions and provider setup
morphz completion	Generate shell completion definitions
morphz version	Print the Morphz version

Top-level command help

Morphz is an S-Expression Cognitive Machine with persistent Context, Sessions, Objectives and a fullscreen terminal UI. The language model is its nondeterministic semantic processor; the Runtime is its deterministic transactional kernel.

Text entered without a subcommand is sent directly to the selected Agent instance.

Usage: morphz [OPTIONS] [PROMPT]... [COMMAND]

Commands:
  exec
          Run one prompt and print the final reply
  resume
          Reattach an existing or recently active Session
  serve
          Start the HTTP/WebSocket runtime and embedded Dashboard
  dashboard
          Start the Dashboard and open it in the default browser
  edge
          Pair and run an outbound Execution Node
  target
          Inspect and administer Execution Targets
  lease
          Inspect and revoke Target capability leases
  execution
          Inspect and control durable physical Execution Jobs
  setup
          Open guided model provider setup
  provider
          Inspect and verify model providers
  model
          Discover or select models
  profile
          Inspect or select configuration profiles
  context
          Inspect persistent Cognitive Contexts
  scheduler
          Inspect authoritative Scheduler state
  session
          Manage Sessions within a Context
  agent
          Manage persistent Agents
  harness
          Install and inspect versioned Harness packages
  objective
          Manage long-lived Objectives
  trajectory
          Export and verify portable Agent Trajectory Bundles
  storage
          Inspect and migrate Runtime storage authority
  experiment
          Inspect and verify explicitly gated experimental features
  job
          Inspect or cancel delegated Sub Agent jobs
  config
          Inspect resolved configuration and provenance
  update
          Update Morphz from a verified GitHub Release
  doctor
          Check storage, workspace, permissions and provider setup
  completion
          Generate shell completion definitions
  version
          Print the Morphz version
  help
          Print this message or the help of the given subcommand(s)

Arguments:
  [PROMPT]...
          Send text directly to the Agent

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration

      --config-file <FILE>
          Load an explicit trusted configuration file

  -p, --profile <NAME>
          Load a named configuration profile

      --provider <ID>
          Override the configured model provider

  -m, --model <MODEL>
          Override the configured model

      --reasoning-effort <LEVEL>
          Set model reasoning effort

          [possible values: default, auto, none, off, low, medium, high, max]

      --agent <ID>
          Select an Agent

      --context <ID>
          Select or mount a Cognitive Context

      --session <ID>
          Reattach an existing Session

      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation

  -s, --sandbox <MODE>
          Set the command sandbox mode

          [possible values: workspace-write, full-access, danger-full-access]

  -a, --approval <MODE>
          Set the approval policy

          [possible values: human, ask, auto, auto-review, never, deny]

      --add-dir <DIR>
          Add an additional readable and writable Workspace directory

      --network[=<BOOL>]
          Allow sandboxed commands to access the network

          [possible values: true, false, 1, 0, yes, no, on, off]

  -c, --set <KEY=VALUE>
          Override one configuration value

      --log-level <FILTER>
          Override the tracing filter

      --theme <THEME>
          Select the TUI color theme

          [possible values: system, mono, iris, cyan, coral, no-color]

      --language <LANGUAGE>
          Select the user-interface language

          [possible values: auto, en, zh-CN]

      --format <FORMAT>
          Select management-command output format

          [possible values: human, json]

      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process

      --tui
          Force the fullscreen terminal UI

      --plain
          Use the classic line-oriented terminal

  -h, --help
          Print help (see a summary with '-h')

  -V, --version
          Print version

Examples:
  morphz
  morphz please help me fix this project
  morphz -- session list
  morphz session list --format=json
  morphz resume --context=context-default

morphz exec

Run one prompt and print the final reply

Run one prompt and print the final reply

Usage: morphz exec [OPTIONS] <PROMPT>...

Arguments:
  <PROMPT>...
          Prompt to send to the Agent

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Examples:
  morphz exec explain this repository
  morphz exec -- --text-that-starts-with-a-dash

morphz resume

Reattach an existing or recently active Session

Reattach a Session without changing its identity. With no ID, resumes the most recently active matching Session.

Usage: morphz resume [OPTIONS] [[SESSION] [PROMPT]]...

Arguments:
  [[SESSION] [PROMPT]]...
          Optional Session ID followed by an optional prompt

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration

      --last
          Resume the most recently active matching Session

      --config-file <FILE>
          Load an explicit trusted configuration file

  -p, --profile <NAME>
          Load a named configuration profile

      --provider <ID>
          Override the configured model provider

  -m, --model <MODEL>
          Override the configured model

      --reasoning-effort <LEVEL>
          Set model reasoning effort

          [possible values: default, auto, none, off, low, medium, high, max]

      --agent <ID>
          Select an Agent

      --context <ID>
          Select or mount a Cognitive Context

      --session <ID>
          Reattach an existing Session

      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation

  -s, --sandbox <MODE>
          Set the command sandbox mode

          [possible values: workspace-write, full-access, danger-full-access]

  -a, --approval <MODE>
          Set the approval policy

          [possible values: human, ask, auto, auto-review, never, deny]

      --add-dir <DIR>
          Add an additional readable and writable Workspace directory

      --network[=<BOOL>]
          Allow sandboxed commands to access the network

          [possible values: true, false, 1, 0, yes, no, on, off]

  -c, --set <KEY=VALUE>
          Override one configuration value

      --log-level <FILTER>
          Override the tracing filter

      --theme <THEME>
          Select the TUI color theme

          [possible values: system, mono, iris, cyan, coral, no-color]

      --language <LANGUAGE>
          Select the user-interface language

          [possible values: auto, en, zh-CN]

      --format <FORMAT>
          Select management-command output format

          [possible values: human, json]

      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process

      --tui
          Force the fullscreen terminal UI

      --plain
          Use the classic line-oriented terminal

  -h, --help
          Print help (see a summary with '-h')

  -V, --version
          Print version

Examples:
  morphz resume
  morphz resume session_123
  morphz resume session_123 continue the task
  morphz resume --context=context-default

morphz serve

Start the HTTP/WebSocket runtime and embedded Dashboard

Start the HTTP/WebSocket runtime and embedded Dashboard. Loopback addresses may run without Dashboard authentication; non-loopback addresses require MORPHZ_DASHBOARD_TOKEN.

Usage: morphz serve [OPTIONS]

Options:
      --bind <ADDR>
          Listen address

  -C, --cwd <DIR>
          Change working directory before loading configuration

      --config-file <FILE>
          Load an explicit trusted configuration file

      --coordination-mesh <SOURCE>
          Join a Coordination Mesh from static:URL,URL or file:PATH

  -p, --profile <NAME>
          Load a named configuration profile

      --provider <ID>
          Override the configured model provider

  -m, --model <MODEL>
          Override the configured model

      --reasoning-effort <LEVEL>
          Set model reasoning effort

          [possible values: default, auto, none, off, low, medium, high, max]

      --agent <ID>
          Select an Agent

      --context <ID>
          Select or mount a Cognitive Context

      --session <ID>
          Reattach an existing Session

      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation

  -s, --sandbox <MODE>
          Set the command sandbox mode

          [possible values: workspace-write, full-access, danger-full-access]

  -a, --approval <MODE>
          Set the approval policy

          [possible values: human, ask, auto, auto-review, never, deny]

      --add-dir <DIR>
          Add an additional readable and writable Workspace directory

      --network[=<BOOL>]
          Allow sandboxed commands to access the network

          [possible values: true, false, 1, 0, yes, no, on, off]

  -c, --set <KEY=VALUE>
          Override one configuration value

      --log-level <FILTER>
          Override the tracing filter

      --theme <THEME>
          Select the TUI color theme

          [possible values: system, mono, iris, cyan, coral, no-color]

      --language <LANGUAGE>
          Select the user-interface language

          [possible values: auto, en, zh-CN]

      --format <FORMAT>
          Select management-command output format

          [possible values: human, json]

      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process

      --tui
          Force the fullscreen terminal UI

      --plain
          Use the classic line-oriented terminal

  -h, --help
          Print help (see a summary with '-h')

  -V, --version
          Print version

Examples:
  morphz serve
  morphz serve --bind=127.0.0.1:9090
  morphz serve --coordination-mesh=static:http://10.0.0.11:8080,http://10.0.0.12:8080
  morphz serve --coordination-mesh=file:/etc/morphz/mesh.toml
  MORPHZ_DASHBOARD_TOKEN=replace-with-a-secret morphz serve --bind=0.0.0.0:8080

morphz dashboard

Start the Dashboard and open it in the default browser

Start the embedded Dashboard with a cryptographically random temporary authentication token and open its local URL in the default browser.

Usage: morphz dashboard [OPTIONS]

Options:
      --bind <ADDR>
          Listen address

  -C, --cwd <DIR>
          Change working directory before loading configuration

      --config-file <FILE>
          Load an explicit trusted configuration file

      --no-open
          Print the Dashboard URL without opening a browser

  -p, --profile <NAME>
          Load a named configuration profile

      --provider <ID>
          Override the configured model provider

  -m, --model <MODEL>
          Override the configured model

      --reasoning-effort <LEVEL>
          Set model reasoning effort

          [possible values: default, auto, none, off, low, medium, high, max]

      --agent <ID>
          Select an Agent

      --context <ID>
          Select or mount a Cognitive Context

      --session <ID>
          Reattach an existing Session

      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation

  -s, --sandbox <MODE>
          Set the command sandbox mode

          [possible values: workspace-write, full-access, danger-full-access]

  -a, --approval <MODE>
          Set the approval policy

          [possible values: human, ask, auto, auto-review, never, deny]

      --add-dir <DIR>
          Add an additional readable and writable Workspace directory

      --network[=<BOOL>]
          Allow sandboxed commands to access the network

          [possible values: true, false, 1, 0, yes, no, on, off]

  -c, --set <KEY=VALUE>
          Override one configuration value

      --log-level <FILTER>
          Override the tracing filter

      --theme <THEME>
          Select the TUI color theme

          [possible values: system, mono, iris, cyan, coral, no-color]

      --language <LANGUAGE>
          Select the user-interface language

          [possible values: auto, en, zh-CN]

      --format <FORMAT>
          Select management-command output format

          [possible values: human, json]

      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process

      --tui
          Force the fullscreen terminal UI

      --plain
          Use the classic line-oriented terminal

  -h, --help
          Print help (see a summary with '-h')

  -V, --version
          Print version

Examples:
  morphz dashboard
  morphz dashboard --no-open
  morphz dashboard --bind=0.0.0.0:8080

morphz edge

Pair and run an outbound Execution Node

Pair and run an outbound Execution Node

Usage: morphz edge [OPTIONS] [COMMAND]

Commands:
  pairing-code
          Create a short-lived pairing code for the current Principal
  nodes
          List Execution Nodes owned by the current Principal
  revoke
          Revoke one paired Execution Node
  local-leases
          List Provider-local capability leases for this Node
  revoke-local-lease
          Revoke one Provider-local capability lease
  pair
          Pair this device with a Morphz Gateway
  run
          Run the authenticated outbound Edge worker
  rotate-key
          Rotate this Node's device identity key
  status
          Show the paired Node credential and local Target identity
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

morphz target

Inspect and administer Execution Targets

Inspect and administer Execution Targets

Usage: morphz target [OPTIONS] [COMMAND]

Commands:
  list
          List Targets visible to the current Principal
  show
          Inspect one Target
  enable
          Enable one Target
  disable
          Disable one Target
  authorize
          Restrict a Target to an Agent, Context or Thread scope
  authorizations
          List scoped Target authorizations
  revoke-authorization
          Revoke one scoped Target authorization
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

morphz lease

Inspect and revoke Target capability leases

Inspect and revoke Target capability leases

Usage: morphz lease [OPTIONS] [COMMAND]

Commands:
  list
          List active capability leases
  revoke
          Revoke one capability lease
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

morphz execution

Inspect and control durable physical Execution Jobs

Inspect and control durable physical Execution Jobs

Usage: morphz execution [OPTIONS] [COMMAND]

Commands:
  list
          List Execution Jobs
  show
          Inspect one Execution Job
  output
          Read durable stdout/stderr chunks for one Job
  cancel
          Request cancellation of one Job
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

morphz setup

Open guided model provider setup

Start the embedded Dashboard directly in guided model provider setup. Use --tui for the fullscreen terminal wizard on SSH or systems without a browser.

Usage: morphz setup [OPTIONS]

Options:
      --bind <ADDR>
          Dashboard listen address

  -C, --cwd <DIR>
          Change working directory before loading configuration

      --config-file <FILE>
          Load an explicit trusted configuration file

      --no-open
          Print the Setup URL without opening a browser

  -p, --profile <NAME>
          Load a named configuration profile

      --provider <ID>
          Override the configured model provider

  -m, --model <MODEL>
          Override the configured model

      --reasoning-effort <LEVEL>
          Set model reasoning effort

          [possible values: default, auto, none, off, low, medium, high, max]

      --agent <ID>
          Select an Agent

      --context <ID>
          Select or mount a Cognitive Context

      --session <ID>
          Reattach an existing Session

      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation

  -s, --sandbox <MODE>
          Set the command sandbox mode

          [possible values: workspace-write, full-access, danger-full-access]

  -a, --approval <MODE>
          Set the approval policy

          [possible values: human, ask, auto, auto-review, never, deny]

      --add-dir <DIR>
          Add an additional readable and writable Workspace directory

      --network[=<BOOL>]
          Allow sandboxed commands to access the network

          [possible values: true, false, 1, 0, yes, no, on, off]

  -c, --set <KEY=VALUE>
          Override one configuration value

      --log-level <FILTER>
          Override the tracing filter

      --theme <THEME>
          Select the TUI color theme

          [possible values: system, mono, iris, cyan, coral, no-color]

      --language <LANGUAGE>
          Select the user-interface language

          [possible values: auto, en, zh-CN]

      --format <FORMAT>
          Select management-command output format

          [possible values: human, json]

      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process

      --tui
          Force the fullscreen terminal UI

      --plain
          Use the classic line-oriented terminal

  -h, --help
          Print help (see a summary with '-h')

  -V, --version
          Print version

Examples:
  morphz setup
  morphz setup --tui
  morphz setup --no-open --bind=127.0.0.1:9090

morphz provider

Inspect and verify model providers

Inspect and verify model providers

Usage: morphz provider [OPTIONS] [COMMAND]

Commands:
  list
          List catalog and configured providers
  test
          Verify a provider catalog, stream and tool call
  show
          Show one effective Provider Instance
  set
          Validate and persist a Provider Instance TOML file
  account
          Manage provider authentication accounts
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz provider <COMMAND> --help` for command-specific help.

morphz model

Discover or select models

Discover or select models

Usage: morphz model [OPTIONS] [COMMAND]

Commands:
  list
          List models exposed by a provider
  use
          Persist the default provider and model
  refresh
          Refresh and verify the remote catalog for one Model Route
  route
          Manage logical Model Routes
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz model <COMMAND> --help` for command-specific help.

morphz profile

Inspect or select configuration profiles

Inspect or select configuration profiles

Usage: morphz profile [OPTIONS] [COMMAND]

Commands:
  list
          List available profiles
  show
          Show the resolved contents of a profile
  use
          Select the default profile
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz profile <COMMAND> --help` for command-specific help.

morphz context

Inspect persistent Cognitive Contexts

Inspect persistent Cognitive Contexts

Usage: morphz context [OPTIONS] [COMMAND]

Commands:
  list
          List Cognitive Contexts
  show
          Show one Cognitive Context
  status
          Show Context state, Sessions and active work
  audit
          Verify the Context Mind projection by replaying Events
  recall-index
          Inspect or rebuild the derived lexical Recall index
  recall
          Search Context memory or traverse one Mind Frame lineage
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz context <COMMAND> --help` for command-specific help.

morphz scheduler

Inspect authoritative Scheduler state

Inspect authoritative Scheduler state

Usage: morphz scheduler [OPTIONS] [COMMAND]

Commands:
  show
          Show Threads, activations, jobs, approvals and schedules
  thread
          Inspect and control one durable Thread
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz scheduler <COMMAND> --help` for command-specific help.

morphz session

Manage Sessions within a Context

Manage Sessions within a Context

Usage: morphz session [OPTIONS] [COMMAND]

Commands:
  list
          List Sessions
  show
          Show one Session
  create
          Create a Session within a selected Context
  resume
          Reattach an existing or recently active Session
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz session <COMMAND> --help` for command-specific help.

morphz agent

Manage persistent Agents

Manage persistent Agents

Usage: morphz agent [OPTIONS] [COMMAND]

Commands:
  list
          List Agents
  show
          Show one Agent
  create
          Create an Agent with a Root Context and initial Session
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz agent <COMMAND> --help` for command-specific help.

morphz harness

Install and inspect versioned Harness packages

Install and inspect versioned Harness packages

Usage: morphz harness [OPTIONS] [COMMAND]

Commands:
  list
          List installed Harness versions
  show
          Show one exact installed Harness version
  install
          Validate and install a .hns file or directory
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz harness <COMMAND> --help` for command-specific help.

morphz objective

Manage long-lived Objectives

Manage long-lived Objectives

Usage: morphz objective [OPTIONS] [COMMAND]

Commands:
  list
          List Objectives in a Context
  show
          Show one Objective
  create
          Create and run a long-lived Objective
  edit
          Replace an Objective goal using revision fencing
  pause
          Pause an Objective
  resume
          Resume an Objective
  cancel
          Cancel an Objective
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz objective <COMMAND> --help` for command-specific help.

morphz trajectory

Export and verify portable Agent Trajectory Bundles

Export and verify portable Agent Trajectory Bundles

Usage: morphz trajectory [OPTIONS] [COMMAND]

Commands:
  export
          Export authoritative Runtime facts as an Agent Trajectory Bundle
  verify
          Validate an untrusted Agent Trajectory Bundle
  episode
          Derive a permission-checked training Episode
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

morphz storage

Inspect and migrate Runtime storage authority

Inspect and migrate Runtime storage authority

Usage: morphz storage [OPTIONS] [COMMAND]

Commands:
  migrate-cognitive-store
          Explicitly synchronize cognitive state into the selected Store
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Runtime startup never migrates cognitive state implicitly.

morphz experiment

Inspect and verify explicitly gated experimental features

Inspect and verify explicitly gated experimental features

Usage: morphz experiment [OPTIONS] [COMMAND]

Commands:
  list
          List compiled and enabled experimental features
  check
          Require one experiment to be compiled and enabled
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Experiments are disabled by default and carry no stability promise.

morphz job

Inspect or cancel delegated Sub Agent jobs

Inspect or cancel delegated Sub Agent jobs

Usage: morphz job [OPTIONS] [COMMAND]

Commands:
  list
          List delegated jobs
  cancel
          Cancel a delegated job and its descendants
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz job <COMMAND> --help` for command-specific help.

morphz config

Inspect resolved configuration and provenance

Inspect resolved configuration and provenance

Usage: morphz config [OPTIONS] [COMMAND]

Commands:
  show
          Print the resolved configuration
  check
          Validate all loaded configuration layers
  path
          List loaded configuration files in precedence order
  explain
          Explain the source of every resolved value
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Run `morphz config <COMMAND> --help` for command-specific help.

morphz update

Update Morphz from a verified GitHub Release

Update Morphz from a verified GitHub Release

Usage: morphz update [OPTIONS] [COMMAND]

Commands:
  status
          Check the latest release without changing any files
  rollback
          Restore the version retained by the last update
  help
          Print this message or the help of the given subcommand(s)

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --to <VERSION>
          Install an exact released version instead of latest
      --allow-downgrade
          Allow --to to install an older release
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Examples:
  morphz update status
  morphz update
  morphz update --to 0.2.0
  morphz update rollback

morphz doctor

Check storage, workspace, permissions and provider setup

Check storage, workspace, permissions and provider setup

Usage: morphz doctor [OPTIONS]

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Example:
  morphz doctor

morphz completion

Generate shell completion definitions

Generate shell completion definitions

Usage: morphz completion [OPTIONS] <SHELL>

Arguments:
  <SHELL>
          Shell to generate completions for [possible values: bash, elvish, fish, powershell, zsh]

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Example:
  morphz completion zsh > ~/.zfunc/_morphz

morphz version

Print the Morphz version

Print the Morphz version

Usage: morphz version [OPTIONS]

Options:
  -C, --cwd <DIR>
          Change working directory before loading configuration
      --config-file <FILE>
          Load an explicit trusted configuration file
  -p, --profile <NAME>
          Load a named configuration profile
      --provider <ID>
          Override the configured model provider
  -m, --model <MODEL>
          Override the configured model
      --reasoning-effort <LEVEL>
          Set model reasoning effort [possible values: default, auto, none, off, low, medium, high, max]
      --agent <ID>
          Select an Agent
      --context <ID>
          Select or mount a Cognitive Context
      --session <ID>
          Reattach an existing Session
      --harness <ID@VERSION>
          Select an exact installed Harness for the initial Evaluation
  -s, --sandbox <MODE>
          Set the command sandbox mode [possible values: workspace-write, full-access, danger-full-access]
  -a, --approval <MODE>
          Set the approval policy [possible values: human, ask, auto, auto-review, never, deny]
      --add-dir <DIR>
          Add an additional readable and writable Workspace directory
      --network[=<BOOL>]
          Allow sandboxed commands to access the network [possible values: true, false, 1, 0, yes, no, on, off]
  -c, --set <KEY=VALUE>
          Override one configuration value
      --log-level <FILTER>
          Override the tracing filter
      --theme <THEME>
          Select the TUI color theme [possible values: system, mono, iris, cyan, coral, no-color]
      --language <LANGUAGE>
          Select the user-interface language [possible values: auto, en, zh-CN]
      --format <FORMAT>
          Select management-command output format [possible values: human, json]
      --enable-experimental <FEATURE>
          Enable one compiled experimental feature for this process
      --tui
          Force the fullscreen terminal UI
      --plain
          Use the classic line-oriented terminal
  -h, --help
          Print help
  -V, --version
          Print version

Example:
  morphz version

Discover deeper help

The current binary remains authoritative for every nested flag. Use morphz help <COMMAND> or append --help to any command path. Automation should consume --format=json and stable IDs instead of parsing human tables or translated text.