Skip to content
LogoLogo

too

Toolang — a language and runtime for agents and humans. (0.4.0)

GroupCommands
Agentnew, clone, remove, info, ls, cd
Cappsyche, skill, service, prompt
Runtimeserve, start, stop, hub
Workchat, chore, talk, convo, workspace
Inspectioncaps, tools, models, providers, inspect, top, ps
Scriptinit, run
Runsteer, cancel, rerun
Threadfork, rewind
Plugincatalogs, adapters, toolsets, sandboxes
Languageparse, highlight, fmt
Helpmore, help

Usage

Usage: too [OPTIONS] [AGENT] <COMMAND> [ARGUMENTS]
 
Agent commands:
  new        Create an agent
  clone      Clone an agent
  remove     Remove an agent
  info       Show agent information
  ls         List agents and their status
  cd         Enter an agent home
 
Cap commands:
  psyche     Manage psyche caps
  skill      Manage skill caps
  service    Manage service caps
  prompt     Manage prompt caps
 
Runtime commands:
  serve      Run an agent in the foreground
  start      Start an agent in the background
  stop       Stop an agent
  hub        Manage the hub service
 
Work commands:
  chat       Chat directly with an agent
  chore      Manage one-off and scheduled tasks
  talk       Have conversations with agents
  convo      Manage conversations and rules
  workspace  Manage agent workspaces
 
Inspection commands:
  caps       List available caps
  tools      List available tools
  models     List available models
  providers  List available model providers
  inspect    Inspect agent runs
  top        Monitor agent runs
  ps         Show agents and hub status
 
Script commands:
  init       Initialize Toolang in a directory
  run        Execute a .too file
 
Options:
  -r, --root <PATH>  Use a custom Toolang root
  -V, --version      Show version and exit
  -h, --help         Show this message and exit
 
Run 'too more' to see additional commands.
Run 'too help' to see option help topics.

too new

Create an agent.

Usage: too new [OPTIONS] <NAME>
 
Arguments:
  * NAME  New agent name
 
Options:
  -h, --help  Show this message and exit

too clone

Clone an agent.

Usage: too clone [OPTIONS] <SOURCE> [NAME]
 
Arguments:
  * SOURCE  Agent name, ref or URL
    NAME    New agent name; required for resident agents
 
Options:
  -h, --help  Show this message and exit

too remove

Remove an agent.

Usage: too remove [OPTIONS] <NAME>
 
Arguments:
  * NAME  Agent name
 
Options:
  -h, --help  Show this message and exit

too info

Show agent information.

Usage: too info <AGENT> [OPTIONS]
 
Arguments:
  * AGENT  Agent name, path, ref, URL or instance
 
Options:
      --catalog <PATH>  Use this model catalog
  -h, --help            Show this message and exit

too ls

List agents and their status.

Usage: too ls [OPTIONS]
 
Options:
  -h, --help  Show this message and exit

too cd

Enter an agent home.

Usage: too cd [AGENT] [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to open a shell in the Toolang root.

too psyche

Manage psyche caps.

Usage: too [AGENT] psyche [OPTIONS] <COMMAND> [ARGUMENTS]
 
Commands:
  list    List psyches
  new     Create a psyche
  edit    Edit a psyche
  delete  Delete a psyche
  add     Add a remote psyche reference
  remove  Remove a psyche reference
 
Options:
  -h, --help  Show this message and exit

too psyche list

List psyches.

Usage: too [AGENT] psyche list [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
  -q, --query <QUERY>  TQ query over cap (kind/name), source, and tags; repeat
                       for union
  -a, --all            Include allow-excluded caps
      --json           Output caps as JSON
      --human          Display a table (default)
  -h, --help           Show this message and exit
 
Omit AGENT to list caps from the Toolang root.

too psyche new

Create a psyche.

Usage: too [AGENT] psyche new [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Psyche name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too psyche edit

Edit a psyche.

Usage: too [AGENT] psyche edit [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Psyche name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too psyche delete

Delete a psyche.

Usage: too [AGENT] psyche delete [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Psyche name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too psyche add

Add a remote psyche reference.

Usage: too [AGENT] psyche add [OPTIONS] <SOURCE>
 
Arguments:
    AGENT   Agent name
  * SOURCE  Psyche ref or URL
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too psyche remove

Remove a psyche reference.

Usage: too [AGENT] psyche remove [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Psyche name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too skill

Manage skill caps.

Usage: too [AGENT] skill [OPTIONS] <COMMAND> [ARGUMENTS]
 
Commands:
  list    List skills
  new     Create a skill
  edit    Edit a skill
  delete  Delete a skill
  add     Add a remote skill reference
  remove  Remove a skill reference
 
Options:
  -h, --help  Show this message and exit

too skill list

List skills.

Usage: too [AGENT] skill list [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
  -q, --query <QUERY>  TQ query over cap (kind/name), source, and tags; repeat
                       for union
  -a, --all            Include allow-excluded caps
      --json           Output caps as JSON
      --human          Display a table (default)
  -h, --help           Show this message and exit
 
Omit AGENT to list caps from the Toolang root.

too skill new

Create a skill.

Usage: too [AGENT] skill new [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Skill name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too skill edit

Edit a skill.

Usage: too [AGENT] skill edit [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Skill name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too skill delete

Delete a skill.

Usage: too [AGENT] skill delete [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Skill name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too skill add

Add a remote skill reference.

Usage: too [AGENT] skill add [OPTIONS] <SOURCE>
 
Arguments:
    AGENT   Agent name
  * SOURCE  Skill ref or URL
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too skill remove

Remove a skill reference.

Usage: too [AGENT] skill remove [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Skill name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too service

Manage service caps.

Usage: too [AGENT] service [OPTIONS] <COMMAND> [ARGUMENTS]
 
Commands:
  list    List services
  new     Create a service
  edit    Edit a service
  delete  Delete a service
  add     Add a remote service reference
  remove  Remove a service reference
 
Options:
  -h, --help  Show this message and exit

too service list

List services.

Usage: too [AGENT] service list [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
  -q, --query <QUERY>  TQ query over cap (kind/name), source, and tags; repeat
                       for union
  -a, --all            Include allow-excluded caps
      --json           Output caps as JSON
      --human          Display a table (default)
  -h, --help           Show this message and exit
 
Omit AGENT to list caps from the Toolang root.

too service new

Create a service.

Usage: too [AGENT] service new [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Service name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too service edit

Edit a service.

Usage: too [AGENT] service edit [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Service name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too service delete

Delete a service.

Usage: too [AGENT] service delete [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Service name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too service add

Add a remote service reference.

Usage: too [AGENT] service add [OPTIONS] <SOURCE>
 
Arguments:
    AGENT   Agent name
  * SOURCE  Service ref or URL
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too service remove

Remove a service reference.

Usage: too [AGENT] service remove [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Service name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too prompt

Manage prompt caps.

Usage: too [AGENT] prompt [OPTIONS] <COMMAND> [ARGUMENTS]
 
Commands:
  list    List prompts
  new     Create a prompt
  edit    Edit a prompt
  delete  Delete a prompt
  add     Add a remote prompt reference
  remove  Remove a prompt reference
 
Options:
  -h, --help  Show this message and exit

too prompt list

List prompts.

Usage: too [AGENT] prompt list [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
  -q, --query <QUERY>  TQ query over cap (kind/name), source, and tags; repeat
                       for union
  -a, --all            Include allow-excluded caps
      --json           Output caps as JSON
      --human          Display a table (default)
  -h, --help           Show this message and exit
 
Omit AGENT to list caps from the Toolang root.

too prompt new

Create a prompt.

Usage: too [AGENT] prompt new [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Prompt name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too prompt edit

Edit a prompt.

Usage: too [AGENT] prompt edit [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Prompt name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too prompt delete

Delete a prompt.

Usage: too [AGENT] prompt delete [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Prompt name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too prompt add

Add a remote prompt reference.

Usage: too [AGENT] prompt add [OPTIONS] <SOURCE>
 
Arguments:
    AGENT   Agent name
  * SOURCE  Prompt ref or URL
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too prompt remove

Remove a prompt reference.

Usage: too [AGENT] prompt remove [OPTIONS] <NAME>
 
Arguments:
    AGENT  Agent name
  * NAME   Prompt name
 
Options:
  -h, --help  Show this message and exit
 
Omit AGENT to modify caps in the Toolang root.

too serve

Run an agent in the foreground.

Usage: too serve <AGENT> [OPTIONS]
 
Arguments:
  * AGENT  Agent name, ref, URL or instance
 
Options:
      --sandbox <SANDBOX>            Run in this sandbox; defaults to config,
                                     last successful selection, or host
      --allow <KEY>=<VALUE>          Select allowed resources; see 'too help
                                     allow'
      --limit <KEY>=<VALUE>          Set run limits; see 'too help limit'
      --default <KEY>=<VALUE>        Set runtime or Chat session defaults; see
                                     'too help default'
      --compact-model <KEY>=<VALUE>  Set the compaction model; see 'too help
                                     model'
      --enable <DRIVERS>             Enable only these drivers, replacing
                                     defaults: chat,chore,talk or none
      --catalog <PATH>               Use this model catalog
      --host <HOST>                  Bind the agent API to this host  [default:
                                     127.0.0.1]
      --port <PORT>                  Bind the agent API to this port
  -w, --workspace [NAME=]<DIR>       Add a temporary workspace (repeatable),
                                     e.g. 'repo=.'
  -d, --workdir [NAME=]<DIR>|<URI>   Set the working directory (once per
                                     invocation), e.g. 'repo://src'. A directory
                                     also adds a temporary workspace
      --dev [PATH]                   Use a local Toolang wheel  [bare: .]
  -h, --help                         Show this message and exit

too start

Start an agent in the background.

Usage: too start <AGENT> [OPTIONS]
 
Arguments:
  * AGENT  Agent name, ref, URL or instance
 
Options:
      --sandbox <SANDBOX>            Run in this sandbox; defaults to config,
                                     last successful selection, or host
      --allow <KEY>=<VALUE>          Select allowed resources; see 'too help
                                     allow'
      --limit <KEY>=<VALUE>          Set run limits; see 'too help limit'
      --default <KEY>=<VALUE>        Set runtime or Chat session defaults; see
                                     'too help default'
      --compact-model <KEY>=<VALUE>  Set the compaction model; see 'too help
                                     model'
      --enable <DRIVERS>             Enable only these drivers, replacing
                                     defaults: chat,chore,talk or none
      --catalog <PATH>               Use this model catalog
      --host <HOST>                  Bind the agent API to this host  [default:
                                     127.0.0.1]
      --port <PORT>                  Bind the agent API to this port
      --dev [PATH]                   Use a local Toolang wheel  [bare: .]
  -h, --help                         Show this message and exit

too stop

Stop an agent.

Usage: too stop <AGENT> [OPTIONS]
 
Arguments:
  * AGENT  Agent name, ref, URL or instance
 
Options:
      --force / --no-force  Force-stop when graceful shutdown does not complete
                            [default: no-force]
  -h, --help                Show this message and exit

too hub

Manage the hub service.

Usage: too hub [COMMAND]
 
Commands:
  status  Show hub status
  start   Start hub in the background
  stop    Stop hub
  serve   Run hub in the foreground
 
Options:
  -h, --help  Show this message and exit

too hub status

Show hub status.

Usage: too hub status [OPTIONS]
 
Options:
  -h, --help  Show this message and exit

too hub start

Start hub in the background.

Usage: too hub start [OPTIONS]
 
Options:
      --port <PORT>  Bind the hub API to this port  [1<=x<=65535]
  -h, --help         Show this message and exit

too hub stop

Stop hub.

Usage: too hub stop [OPTIONS]
 
Options:
      --force / --no-force  Force-stop when graceful shutdown does not complete
                            [default: no-force]
  -h, --help                Show this message and exit

too hub serve

Run hub in the foreground.

Usage: too hub serve [OPTIONS]
 
Options:
      --port <PORT>  Bind the hub API to this port  [1<=x<=65535]
  -h, --help         Show this message and exit

too chat

Chat directly with an agent.

Usage: too chat <AGENT> [OPTIONS]
 
Arguments:
  * AGENT  Agent name, path, ref, URL or instance
 
Options:
  -t, --thread [THREAD]              Continue a thread by ID  [bare: latest
                                     thread]
      --catalog <PATH>               Use this model catalog
      --sandbox <SANDBOX>            Execute the session in this sandbox
      --allow <KEY>=<VALUE>          Select allowed resources; see 'too help
                                     allow'
      --limit <KEY>=<VALUE>          Set run limits; see 'too help limit'
      --default <KEY>=<VALUE>        Set runtime or Chat session defaults; see
                                     'too help default'
      --compact-model <KEY>=<VALUE>  Set the compaction model; see 'too help
                                     model'
      --enable <DRIVERS>             Enable only these drivers, replacing
                                     defaults: chat,chore,talk or none
  -w, --workspace [NAME=]<DIR>       Add a temporary workspace (repeatable),
                                     e.g. 'repo=.'
  -d, --workdir [NAME=]<DIR>|<URI>   Set the working directory (once per
                                     invocation), e.g. 'repo://src'. A directory
                                     also adds a temporary workspace
      --dev [PATH]                   Use a local Toolang wheel  [bare: .]
  -h, --help                         Show this message and exit

too chore

Manage one-off and scheduled tasks.

Usage: too <AGENT> chore [OPTIONS] <COMMAND> [ARGUMENTS]
 
Commands:
  list    List chores and status
  new     Create a chore
  edit    Edit a chore
  delete  Delete a chore; keep history
  run     Queue one manual chore run
  cancel  Cancel the current chore occurrence; keep future scheduled work
 
Options:
  -h, --help  Show this message and exit

too chore list

List chores and status.

Usage: too <AGENT> chore list [OPTIONS]
 
Arguments:
  * AGENT  Agent name
 
Options:
  -h, --help  Show this message and exit

too chore new

Create a chore.

Usage: too <AGENT> chore new [OPTIONS] [NAME]
 
Arguments:
  * AGENT  Agent name
    NAME   Chore filename without .md; generated when omitted
 
Options:
  -h, --help  Show this message and exit

too chore edit

Edit a chore.

Usage: too <AGENT> chore edit [OPTIONS] <CHORE>
 
Arguments:
  * AGENT  Agent name
  * CHORE  Chore name, ref, or ID
 
Options:
  -h, --help  Show this message and exit

too chore delete

Delete a chore; keep history.

Usage: too <AGENT> chore delete [OPTIONS] <CHORE>
 
Arguments:
  * AGENT  Agent name
  * CHORE  Chore name, ref, or ID
 
Options:
  -h, --help  Show this message and exit

too chore run

Queue one manual chore run.

Usage: too <AGENT> chore run [OPTIONS] <CHORE>
 
Arguments:
  * AGENT  Running agent name
  * CHORE  Chore name, ref, or ID
 
Options:
  -h, --help  Show this message and exit

too chore cancel

Cancel the current chore occurrence; keep future scheduled work.

Usage: too <AGENT> chore cancel [OPTIONS] <CHORE>
 
Arguments:
  * AGENT  Running agent name
  * CHORE  Chore name, ref, or ID
 
Options:
  -h, --help  Show this message and exit

too talk

Have conversations with agents.

Usage: too talk [OPTIONS] [CONVO] [MESSAGE...]
 
Arguments:
  CONVO    Conversation ID, agent name, comma-separated agent names, or all
  MESSAGE  Message to send; omit to open the conversation
 
Options:
  -h, --help  Show this message and exit
 
Examples:
  too talk gc_abcd1234          Open a conversation by ID
  too talk gc_abcd1234 "Hello"  Send a message and exit
  too talk alice                Open a conversation with an agent
  too talk all                  Open the conversation with all agents
  too talk alice,bob            View a conversation between agents
  too convo list                List conversations

too convo

Manage conversations and rules.

Usage: too convo [OPTIONS] <COMMAND> [ARGUMENTS]
 
Commands:
  list    List conversations
  create  Create a group conversation
  join    Join a group conversation
  leave   Leave a group conversation
  rules   Retrieve or update rules
 
Options:
  -h, --help  Show this message and exit

too convo list

List conversations.

Usage: too convo list [OPTIONS]
 
Options:
  -h, --help  Show this message and exit

too convo create

Create a group conversation.

Usage: too convo create [OPTIONS] [NAME]
 
Arguments:
  NAME  Group name, 1–128 characters
 
Options:
  -h, --help  Show this message and exit

too convo join

Join a group conversation.

Usage: too convo join [OPTIONS] <CONVO>
 
Arguments:
  * CONVO  Conversation ID starting with dm_ or gc_
 
Options:
  -h, --help  Show this message and exit

too convo leave

Leave a group conversation.

Usage: too convo leave [OPTIONS] <CONVO>
 
Arguments:
  * CONVO  Conversation ID starting with dm_ or gc_
 
Options:
  -h, --help  Show this message and exit

too convo rules

Retrieve or update rules.

Usage: too convo rules [OPTIONS] <CONVO>
 
Arguments:
  * CONVO  Conversation ID starting with dm_ or gc_
 
Options:
  -o, --output <FILE>        Write rules to FILE; - for stdout
  -u, --update [FILE]        Update rules from FILE; omit FILE or use - to read
                             stdin
      --revision <REVISION>  Use the revision returned by rules; required with
                             --update
  -h, --help                 Show this message and exit

too workspace

Manage agent workspaces.

Usage: too <AGENT> workspace [OPTIONS] <COMMAND> [ARGUMENTS]
 
Commands:
  list    List configured workspaces
  add     Add a workspace
  remove  Remove a workspace
 
Options:
  -h, --help  Show this message and exit

too workspace list

List configured workspaces.

Usage: too <AGENT> workspace list [OPTIONS]
 
Arguments:
  * AGENT  Agent name or path
 
Options:
  -h, --help  Show this message and exit

too workspace add

Add a workspace.

Usage: too <AGENT> workspace add [OPTIONS] <PATH>
 
Arguments:
  * AGENT  Agent name or path
  * PATH   Existing directory path
 
Options:
      --name <NAME>  Workspace name, normalized to kebab case
  -h, --help         Show this message and exit

too workspace remove

Remove a workspace.

Usage: too <AGENT> workspace remove [OPTIONS] <NAME>
 
Arguments:
  * AGENT  Agent name or path
  * NAME   Workspace name
 
Options:
  -h, --help  Show this message and exit

too caps

List available caps.

Usage: too [AGENT] caps [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
  -q, --query <QUERY>  TQ query over cap (kind/name), source, and tags; repeat
                       for union
  -a, --all            Include allow-excluded caps
      --json           Output caps as JSON
      --human          Display a table (default)
  -h, --help           Show this message and exit
 
Omit AGENT to list caps from the Toolang root.

too tools

List available tools.

Usage: too [AGENT] tools [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
  -q, --query <QUERY>  TQ query; repeat for union. See --json for fields
  -a, --all            Include internal and allow-excluded tools
      --json           Output tools as JSON
      --human          Display a table (default)
  -h, --help           Show this message and exit
 
Omit AGENT to use settings from the Toolang root.

too models

List available models.

Usage: too [AGENT] models [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
      --catalog <PATH>  Use this model catalog
  -a, --all             Include unready and allow-excluded models
  -q, --query <QUERY>   TQ query; repeat for union. See --json for fields. Costs
                        are per million tokens
      --json            Output models as JSON
      --human           Display a table (default)
  -h, --help            Show this message and exit
 
Omit AGENT to use settings from the Toolang root.

too providers

List available model providers.

Usage: too [AGENT] providers [OPTIONS]
 
Arguments:
  AGENT  Agent name, path, ref or URL
 
Options:
      --catalog <PATH>  Use this model catalog
  -a, --all             Include unready, allow-excluded, and empty providers
      --json            Output providers as JSON
      --human           Display a table (default)
  -h, --help            Show this message and exit
 
Omit AGENT to use settings from the Toolang root.

too inspect

Inspect agent runs.

Usage: too <AGENT> inspect [OPTIONS] <QUERY>...
 
Arguments:
  * AGENT  Agent name, path, ref, URL or instance
  * QUERY  Record ID or collection, with optional relations and a view
 
Options:
      --human  Output human-readable text (default)
      --json   Output JSON (resolved value for run output)
  -h, --help   Show this message and exit
 
Records:
  threads   List threads
  runs      List runs
  controls  List controls
  <ID>      Select a thread, control, run or step ID, or a field path
 
Relations:
  <THREAD> runs  List child runs
  <RUN> steps    List child steps
  <STEP> runs    List child runs
  <STEP> steps   List body steps of a loop
 
Views:
  <RUN> output  Resolved run result
  <RUN> tree    Stored execution tree
  <STEP> call   Historical call owned by the step
 
Examples:
  too alice inspect runs                 List runs
  too alice inspect run_abcd1234 steps   List steps in a run
  too alice inspect run_abcd1234 output  Show a run result

too top

Monitor agent runs.

Usage: too [AGENT] top [OPTIONS]
 
Arguments:
  AGENT  Agent name
 
Options:
      --period <1h|4h|8h>   Rolling usage period  [default: 1h]
      --recent <1m|5m|15m>  Recently completed root Runs  [default: 5m]
      --refresh <SECONDS>   Refresh interval in seconds  [default: 1]
      --once                Print one snapshot
      --json                Print one JSON snapshot
  -h, --help                Show this message and exit
 
Omit AGENT to monitor all agents in the Toolang root.

too ps

Show agents and hub status.

Usage: too ps [OPTIONS]
 
Options:
  -h, --help  Show this message and exit

too init

Initialize Toolang in a directory.

Usage: too init [OPTIONS] <DIR>
 
Arguments:
  * DIR  Directory to set up for Toolang
 
Options:
  -h, --help  Show this message and exit

too run

Execute a .too file.

Usage: too run <FILE> [RUNNABLE] [OPTIONS] [ARGUMENTS] [INPUT...]
 
Arguments:
  * FILE       Path to a .too file
    RUNNABLE   Runnable name
    ARGUMENTS  Named runnable inputs
    INPUT      Primary runnable input
 
Options:
  -h, --help  Show this message and exit
 
Add --help after FILE or RUNNABLE for details.

too steer

Steer an active run.

Usage: too <AGENT> steer [OPTIONS] <RUN> <MESSAGE>
 
Arguments:
  * AGENT    Running agent name, path, ref, URL or instance
  * RUN      Run ID or thread ID; a thread selects its active run
  * MESSAGE  Instruction to steer the run
 
Options:
  -h, --help  Show this message and exit

too cancel

Cancel an active run.

Usage: too <AGENT> cancel [OPTIONS] <RUN>
 
Arguments:
  * AGENT  Running agent name, path, ref, URL or instance
  * RUN    Run ID or thread ID; a thread selects its active run
 
Options:
  -h, --help  Show this message and exit

too rerun

Rerun an earlier run as a new one.

Usage: too <AGENT> rerun [OPTIONS] <RUN>
 
Arguments:
  * AGENT  Agent name, path, ref, URL or instance
  * RUN    Run ID or thread ID; a thread selects its latest visible root run
 
Options:
      --catalog <PATH>       Use this model catalog
      --sandbox <SANDBOX>    Execute the new run in this sandbox
      --allow <KEY>=<VALUE>  Select allowed resources; see 'too help allow'
      --limit <KEY>=<VALUE>  Set run limits; see 'too help limit'
      --model <KEY>=<VALUE>  Set model settings for this run; see 'too help
                             model'
      --dev [PATH]           Use a local Toolang wheel  [bare: .]
  -h, --help                 Show this message and exit

too fork

Fork a thread from an earlier run.

Usage: too <AGENT> fork [OPTIONS] <RUN>
 
Arguments:
  * AGENT  Agent name, path, ref, URL or instance
  * RUN    Run ID to fork through; a thread ID selects its latest root run
 
Options:
      --chat  Open chat on the forked thread
  -h, --help  Show this message and exit

too rewind

Rewind a thread to an earlier run.

Usage: too <AGENT> rewind [OPTIONS] <RUN>
 
Arguments:
  * AGENT  Agent name, path, ref, URL or instance
  * RUN    Run ID to rewind before; a thread ID selects its latest root run
 
Options:
      --chat  Open chat on the rewound thread
  -h, --help  Show this message and exit

too catalogs

List installed model catalogs.

Usage: too catalogs [OPTIONS]
 
Options:
  -h, --help  Show this message and exit

too adapters

List installed model adapters.

Usage: too adapters [OPTIONS]
 
Options:
  -h, --help  Show this message and exit

too toolsets

List installed toolsets.

Usage: too toolsets [OPTIONS]
 
Options:
  -a, --all   Include internal toolsets
  -h, --help  Show this message and exit

too sandboxes

List installed sandboxes.

Usage: too sandboxes [OPTIONS]
 
Options:
  -h, --help  Show this message and exit

too parse

Parse .too source.

Usage: too parse [OPTIONS] <PATH>...
 
Arguments:
  * PATH  Path to a .too file or - for stdin; --check also accepts directories
 
Options:
      --check                  Validate files or directories without printing an
                               AST
      --ast                    Show the semantic AST (default)
      --cst                    Show the raw concrete syntax tree
      --json                   Emit JSON instead of S-expression
      --compact                Emit compact JSON (implies --json)
      --stdin-filepath <PATH>  Path label for stdin
  -h, --help                   Show this message and exit

too highlight

Highlight .too source.

Usage: too highlight [OPTIONS] <FILE>
 
Arguments:
  * FILE  Path to a .too file, or - for stdin
 
Options:
      --color <auto|always|never>  Terminal color policy  [default: (auto)]
      --html                       Export standalone highlighted HTML
      --stdin-filepath <PATH>      Path label for stdin
  -h, --help                       Show this message and exit

too fmt

Format .too source.

Usage: too fmt [OPTIONS] [PATH...]
 
Arguments:
  PATH  File or directory paths to format
 
Options:
      --check                      Exit non-zero if any file is not formatted
      --tab-size <SPACES>          Number of spaces per indentation level
                                   [default: 2]
      --stdin-filepath <PATH>      Path label for stdin
      --stdout                     Print formatted source without writing files
      --highlight                  Print highlighted formatted source without
                                   writing files
      --color <auto|always|never>  Terminal color policy  [default: (auto)]
      --html                       Export standalone highlighted HTML
  -h, --help                       Show this message and exit

too more

Run commands:
  steer      Steer an active run
  cancel     Cancel an active run
  rerun      Rerun an earlier run as a new one
 
Thread commands:
  fork       Fork a thread from an earlier run
  rewind     Rewind a thread to an earlier run
 
Plugin commands:
  catalogs   List installed model catalogs
  adapters   List installed model adapters
  toolsets   List installed toolsets
  sandboxes  List installed sandboxes
 
Language commands:
  parse      Parse .too source
  highlight  Highlight .too source
  fmt        Format .too source
 
Run 'too COMMAND --help' for details.

too help

Read shared option syntax.

Usage: too help [OPTIONS] [TOPIC]
 
Topics:
  allow    Select allowed resources
  limit    Set run limits
  default  Set runtime or Chat session defaults
  model    Select models and set model parameters
 
Options:
  -h, --help  Show this message and exit
 
Run 'too help TOPIC' for details.
Use 'too COMMAND --help' for command-specific options.

too help allow

--allow <KEY>=<VALUE>
 
Select allowed resources.
Repeat a key to combine TQ queries by union.
With all or none, that key may appear only once.
Quote values containing spaces.
Existing policy restrictions still apply.
 
models=QUERY|all|none
  Select allowed models.
 
  Possible values:
    QUERY  Match models with a TQ query
    all    Select all models
    none   Select no models
 
tools=QUERY|all|none
  Select allowed tools.
 
  Possible values:
    QUERY  Match tools with a TQ query
    all    Select all tools
    none   Select no tools
 
psyches=QUERY|all|none
  Select allowed psyches.
 
  Possible values:
    QUERY  Match psyches with a TQ query
    all    Select all psyches
    none   Select no psyches
 
skills=QUERY|all|none
  Select allowed skills.
 
  Possible values:
    QUERY  Match skills with a TQ query
    all    Select all skills
    none   Select no skills
 
services=QUERY|all|none
  Select allowed services.
 
  Possible values:
    QUERY  Match services with a TQ query
    all    Select all services
    none   Select no services
 
prompts=QUERY|all|none
  Select allowed prompts.
 
  Possible values:
    QUERY  Match prompts with a TQ query
    all    Select all prompts
    none   Select no prompts
 
Examples:
  Allow no tools
    too gus chat --allow tools=none
 
  Select OpenAI models
    too gus chat --allow 'models=openai/*'

too help limit

--limit <KEY>=<VALUE>
 
Set run limits.
Each key may appear once.
 
agic_model_calls=INTEGER|none
  Maximum model calls per agic.
 
  Possible values:
    INTEGER  Non-negative integer
    none     Unset the limit
 
agic_tool_calls=INTEGER|none
  Maximum tool calls per agic.
 
  Possible values:
    INTEGER  Non-negative integer
    none     Unset the limit
 
tokens=INTEGER|none
  Total input and output tokens per run tree.
 
  Possible values:
    INTEGER  Non-negative integer
    none     Unset the limit
 
cost=DECIMAL|none
  Total cost in USD per run tree.
 
  Possible values:
    DECIMAL  Non-negative decimal up to 999999999.999999; rounded to six decimal
             places
    none     Unset the limit
 
time=INTEGER|none
  Elapsed time in seconds per run tree.
 
  Possible values:
    INTEGER  Non-negative integer
    none     Unset the limit
 
Examples:
  Limit tokens and cost
    too gus chat --limit tokens=100000 \
      --limit cost=1.5
 
  Unset the time limit
    too gus chat --limit time=none

too help default

--default <KEY>=<VALUE>
 
Set runtime or Chat session defaults.
Each key may appear once. Quote values containing spaces.
 
model=SETTINGS
  Set the runtime or Chat session model; see 'too help model'.
 
runnable=REF|none
  Select the Chat session runnable.
 
  Possible values:
    REF   Runnable ref
    none  Use automatic entry selection
 
Applies to:
  serve, start   model; applies to the runtime instance
  chat           model and runnable; apply to the current session
  Script, rerun  Use --model for one run
 
  Chat defaults do not reconfigure the runtime or other sessions.
 
Configuration:
  Use [model] for model configuration.
  TOOLANG_DEFAULT_MODEL supplies runtime model defaults.
 
Examples:
  Set the session model
    too gus chat --default \
      'model=openai/gpt-5 effort=high'
 
  Select the session runnable
    too gus chat \
      --default runnable=flow:review
 
See also:
  too help model  Model value syntax

too help model

Model selection
 
Syntax:
  --model <KEY>=<VALUE>          Script runs or reruns
  --default model=<SETTINGS>     Runtime or Chat session defaults
  --compact-model <KEY>=<VALUE>  Compaction model for a new runtime
 
Model settings:
  A model ref may also be given without ref=. Put a bare ref first.
  Omit the ref to change parameters on an effective model.
  Pass multiple settings in one quoted value.
  Each setting may appear only once.
  --compact-model requires an exact model ref. It also accepts effort and
  max_output settings.
 
  ref=MODEL|default|unset
    Select a model.
 
    Script runs and Chat per-run overrides accept unset for model-free
    execution.
 
    Rerun accepts an exact model ref, default, or parameter overrides.
 
    Possible values:
      MODEL    Exact model ref
      default  Use the inherited model or session baseline
      unset    Clear selection; startup entry fallback still applies
 
  effort=LEVEL|INTEGER|auto
    Set reasoning effort.
 
    Possible values:
      LEVEL    Reasoning level supported by the selected model
      INTEGER  Non-negative token budget
      auto     Clear the reasoning parameter
 
  max_output=INTEGER|auto
    Set the maximum output token count.
 
    Possible values:
      INTEGER  Positive output token count
      auto     Use automatic output sizing
 
Examples:
  Set the model for a script run
    too run demo.too process --model \
      'ref=openai/gpt-5 effort=high'
 
  Set a model with reasoning
    too gus chat --default \
      'model=openai/gpt-5 effort=high'
 
  Set the compaction model
    too gus start --compact-model \
      'ref=openai/gpt-5 max_output=2048'
 
See also:
  too models        List available models
  too help default  Default setting lifetimes