microsoft/TypeAgent

---
layout: docs
title: TypeAgent Command Reference
---

The following list of commands are available in the [TypeAgent](/) system. Some commands may not be available based on the client being used (i.e. @shell commands do not work on the CLI) and or may be dependent on a specific agent (i.e. browser). This information is also available by running `@help` in the client.

## @action - Execute an action

Usage: `@action [--parameters <json>] <schemaName> <actionName>`

### Arguments:

- &lt;schemaName&gt; - Action schema name (type: string)
- &lt;actionName&gt; - Action name (type: string)

### Flags:

- --parameters <json> : Action parameter

## @browser launch hidden - Open a hidden/headless browser instance

Usage: `@browser launch hidden`

## @browser launch standalone - Open a standalone browser instance

Usage: `@browser launch standalone`

## @browser close - Close the new Web Content view

Usage: `@browser close`

## @browser resolver list - List all available URL resolvers

Usage: `@browser resolver list`

## @browser resolver search - Toggle search resolver

Usage: `@browser resolver search`

## @browser resolver keyword - Toggle keyword resolver

Usage: `@browser resolver keyword`

## @browser resolver wikipedia - Toggle Wikipedia resolver

Usage: `@browser resolver wikipedia`

## @browser resolver history - Toggle history resolver

Usage: `@browser resolver history`

## @browser search list - Lists browser agent search providers

Usage: `@browser search list`

## @browser search set - Sets the active search provider

Usage: `@browser search set <provider>`
Arguments:
&lt;provider&gt; - The name of the search provider to set as active. (type: string)

## @browser search show - Shows the details of the selected search provider

Usage: `@browser search show <provider>`
Arguments:
&lt;provider&gt; - The name of the search provider to show details for. (type: string)

## @browser search add - Adds a new search provider

Usage: `@browser search add <provider> <url>`
Arguments:
&lt;provider&gt; - The name of the search provider to add. (type: string)
&lt;url&gt; - The URL of the search provider to add. '%s' will be replaced with the search parameter. (type: string)

## @browser search remove - Removes the selected search provider

Usage: `@browser search remove <provider>`
Arguments:
&lt;provider&gt; - The name of the search provider to remove. (type: string)

## @browser search import - Imports the search providers from the specified browser

Usage: `@browser search import <browser>`
Arguments:
&lt;browser&gt; - The name of the browser to import search providers from: [Edge | Chrome]. (type: string)

## @calendar login - Log into MS Graph to access calendar

Usage: `@calendar login`

## @calendar logout - Log out of MS Graph to access calendar

Usage: `@calendar logout`

## @clear - Clear the console

Usage: `@clear`

## @config schema - Toggle agent schemas

Usage: `@config schema [-f|--priority <string>] [-x|--off <string>] [-r|--reset] [<agentNames>...]`

### Arguments:

- &lt;agentNames&gt; - (optional) enable pattern (type: string)

### Flags:

- --reset -r : reset to default (default: false)
- --off -x &lt;string&gt; : disable pattern
- --priority -f &lt;string&gt; : priority pattern

## @config action - Toggle agent actions

Usage: `@config action [-f|--priority <string>] [-x|--off <string>] [-r|--reset] [<agentNames>...]`

### Arguments:

- &lt;agentNames&gt; - (optional) enable pattern (type: string)

### Flags:

- --reset -r : reset to default (default: false)
- --off -x &lt;string&gt; : disable pattern
- --priority -f &lt;string&gt; : priority pattern

## @config command - Toggle agent commands

Usage: `@config command [-f|--priority <string>] [-x|--off <string>] [-r|--reset] [<agentNames>...]`

### Arguments:

- &lt;agentNames&gt; - (optional) enable pattern (type: string)

### Flags:

- --reset -r : reset to default (default: false)
- --off -x &lt;string&gt; : disable pattern
- --priority -f &lt;string&gt; : priority pattern

## @config agent - Toggle agents

Usage: `@config agent [-f|--priority <string>] [-x|--off <string>] [-r|--reset] [<agentNames>...]`

### Arguments:

- &lt;agentNames&gt; - (optional) enable pattern (type: string)

### Flags:

- --reset -r : reset to default (default: false)
- --off -x &lt;string&gt; : disable pattern
- --priority -f &lt;string&gt; : priority pattern

## @config request - Set the agent that handles natural language requests

Usage: `@config request <appAgentName>`

### Arguments:

- &lt;appAgentName&gt; - Name of the agent (type: string)

## @config translation on - Turn on translation

Usage: `@config translation on`

## @config translation off - Turn off translation

Usage: `@config translation off`

## @config translation model - Set model

Usage: `@config translation model [-r|--reset] [<model>]`

### Arguments:

- &lt;model&gt; - (optional) Model name (type: string)

### Flags:

- --reset, -r : Reset to default model (default: false)

## @config translation multi on - Turn on multiple action translation

Usage: `@config translation multi on`

## @config translation multi off - Turn off multiple action translation

Usage: `@config translation multi off`

## @config translation multi result on - Turn on result id in multiple action

Usage: `@config translation multi result on`

## @config translation multi result off - Turn off result id in multiple action

Usage: `@config translation multi result off`

## @config translation multi pending on - Turn on pending request in multiple action

Usage: `@config translation multi pending on`

## @config translation multi pending off - Turn off pending request in multiple action

Usage: `@config translation multi pending off`

## @config translation switch on - Turn on switch schema

Usage: `@config translation switch on`

## @config translation switch off - Turn off switch schema

Usage: `@config translation switch off`

## @config translation switch fix - Set a fixed schema disable switching

Usage: `@config translation switch fix <schemaName>`

### Arguments:

- &lt;schemaName&gt; - name of the schema (type: string)

## @config translation switch inline on - Turn on inject inline switch

Usage: `@config translation switch inline on`

## @config translation switch inline off - Turn off inject inline switch

Usage: `@config translation switch inline off`

## @config translation switch search on - Turn on search switch

Usage: `@config translation switch search on`

## @config translation switch search off - Turn off search switch

Usage: `@config translation switch search off`

## @config translation switch embedding on - Turn on Use embedding for initial pick of schema

Usage: `@config translation switch embedding on`

## @config translation switch embedding off - Turn off Use embedding for initial pick of schema

Usage: `@config translation switch embedding off`

## @config translation history on - Turn on history

Usage: `@config translation history on`

## @config translation history off - Turn off history

Usage: `@config translation history off`

## @config translation history limit - Set the limit of chat history usage in translation

Usage: `@config translation history limit <limit>`

### Arguments:

- &lt;limit&gt; - Number of actions (type: number)

## @config translation stream on - Turn on streaming translation

Usage: `@config translation stream on`

## @config translation stream off - Turn off streaming translation

Usage: `@config translation stream off`

## @config translation schema generation on - Turn on generated action schema

Usage: `@config translation schema generation on`

## @config translation schema generation off - Turn off generated action schema

Usage: `@config translation schema generation off`

## @config translation schema generation json on - Turn on use generate json schema if model supports it

Usage: `@config translation schema generation json on`

## @config translation schema generation json off - Turn off use generate json schema if model supports it

Usage: `@config translation schema generation json off`

## @config translation schema generation jsonFunc on - Turn on use generate json schema function if model supports it

Usage: `@config translation schema generation jsonFunc on`

## @config translation schema generation jsonFunc off - Turn off use generate json schema function if model supports it

Usage: `@config translation schema generation jsonFunc off`

## @config translation schema optimize on - Turn on schema optimization

Usage: `@config translation schema optimize on`

## @config translation schema optimize off - Turn off schema optimization

Usage: `@config translation schema optimize off`

## @config translation schema optimize actions - Set number of actions to use for initial translation

Usage: `@config translation schema optimize actions <count>`

### Arguments:

- &lt;count&gt; - Number of actions (type: number)

## @config explainer on - Turn on explanation

Usage: `@config explainer on`

## @config explainer off - Turn off explanation

Usage: `@config explainer off`

## @config explainer async on - Turn on asynchronous explanation

Usage: `@config explainer async on`

## @config explainer async off - Turn off asynchronous explanation

Usage: `@config explainer async off`

## @config explainer name - Set explainer

Usage: `@config explainer name <explainerName>`

### Arguments:

- &lt;explainerName&gt; - name of the explainer (type: string)

## @config explainer model - Set model

Usage: `@config explainer model [-r|--reset] [<model>]`

### Arguments:

- &lt;model&gt; - (optional) Model name (type: string)

### Flags:

- --reset -r : Reset to default model (default: false)

## @config explainer filter on - Turn on all explanation filters

Usage: `@config explainer filter on`

## @config explainer filter off - Turn off all explanation filters

Usage: `@config explainer filter off`

## @config explainer filter multiple on - Turn on explanation filter multiple actions

Usage: `@config explainer filter multiple on`

## @config explainer filter multiple off - Turn off explanation filter multiple actions

Usage: `@config explainer filter multiple off`

## @config explainer filter reference on - Turn on all explanation reference filters

Usage: `@config explainer filter reference on`

## @config explainer filter reference off - Turn off all explanation reference filters

Usage: `@config explainer filter reference off`

## @config explainer filter reference value on - Turn on explainer filter reference by value in the request

Usage: `@config explainer filter reference value on`

## @config explainer filter reference value off - Turn off explainer filter reference by value in the request

Usage: `@config explainer filter reference value off`

## @config explainer filter reference list on - Turn on explainer filter reference using word lists

Usage: `@config explainer filter reference list on`

## @config explainer filter reference list off - Turn off explainer filter reference using word lists

Usage: `@config explainer filter reference list off`

## @config explainer filter reference translate on - Turn on explainer filter reference by translate without context

Usage: `@config explainer filter reference translate on`

## @config explainer filter reference translate off - Turn off explainer filter reference by translate without context

Usage: `@config explainer filter reference translate off`

## @config serviceHost off - Turn off Service hosting integration

Usage: `@config serviceHost off`

## @config serviceHost on - Turn on Service hosting integration

Usage: `@config serviceHost on`

## @config dev on - Turn on development mode

Usage: `@config dev on`

## @config dev off - Turn off development mode

Usage: `@config dev off`

## @config log db on - Turn on logging

Usage: `@config log db on`

Enables remote upload of dispatcher telemetry (commands, translations, collision events emitted by `@config collision`) to the existing Cosmos `telemetrydb / dispatcherlogs` collection. Requires a `COSMOSDB_CONNECTION_STRING` environment variable. The DB sink self-disables on auth errors so a stale credential won't spam retries.

## @config log db off - Turn off logging

Usage: `@config log db off`

## @config collision - Show / configure action collision detection

Usage: `@config collision`

Default subcommand renders a status table showing each detection point's `detect` toggle, current strategy, calibration knobs (classifier / topN / threshold / scorer), the priority order, multipleActionBehavior, and telemetry settings. The system ships off by default; opt in per-point during a soft-rollout experiment ([`collision-rollout.md`](../../../ts/docs/architecture/collision-rollout.md) is the playbook).

## @config collision &lt;point&gt; detect on|off - Toggle a detection point

Usage: `@config collision <point> detect on`, `@config collision <point> detect off`

`<point>` is one of `static`, `grammarMatch`, `llmSelect`, `fuzzy`. Persists to the session JSON via `session.updateSettings` and re-applies in-memory immediately — no restart needed.

## @config collision &lt;point&gt; strategy &lt;name&gt; - Set the resolution strategy

Usage: `@config collision <point> strategy <name>`

Allowed names depend on the point:

- `static`: `warn`, `error`
- `grammarMatch` / `llmSelect` / `fuzzy`: `first-match`, `score-rank`, `priority`, `user-clarify`

`first-match` preserves legacy behavior (no actual re-ranking). `score-rank` re-orders by cache match metadata (matchedCount / nonOptionalCount / wildcardCharCount) with priority as a tiebreaker. `priority` sorts purely by `priorityOrder` (or registration order when unset). `user-clarify` synthesizes a clarification action listing all candidates so the user picks.

## @config collision priority [&lt;list&gt;] - Set / show priority order

Usage: `@config collision priority "list,player,calendar"`, `@config collision priority ""` (clear), `@config collision priority` (show)

Comma-separated agent names used by the `priority` resolution strategy (and as a tiebreaker for `score-rank`). Empty falls back to agent registration order.

## @config collision telemetry emit on|off - Toggle the in-memory ring buffer + JSONL append

Usage: `@config collision telemetry emit on`, `@config collision telemetry emit off`

When on, every detected collision is recorded to a 50-entry ring buffer accessible via `@collision events` and appended to `<sessionDir>/collision-events.jsonl`. Remote upload to Cosmos is gated separately by `@config log db on`.

## @config collision telemetry debugLog on|off - Toggle the debug-channel log

Usage: `@config collision telemetry debugLog on`

When on, every detected collision emits a line on the `typeagent:dispatcher:collision` debug channel. Useful for `DEBUG=typeagent:dispatcher:collision` capture without enabling the ring buffer.

## @config collision telemetry experimentId [&lt;id&gt;] - Tag emitted events with an experiment ID

Usage: `@config collision telemetry experimentId E1.2-2026-05-12`, `@config collision telemetry experimentId ""` (clear), `@config collision telemetry experimentId` (show)

Each emitted `CollisionEvent` is stamped with this string. Cosmos queries can then `WHERE experimentId = "E1.2-…"` to slice events for a specific rollout experiment without joining other tables.

## @collision events - Show recent collision events from the ring buffer

Usage: `@collision events [-n &lt;limit&gt;] [-k &lt;kind&gt;]`

### Flags:

- `-n, --limit <number>` - Maximum number of events to show (default 10).
- `-k, --kind <name>` - Filter by detection point: `static`, `grammarMatch`, `llmSelect`, or `fuzzy`.

Renders an HTML table of the most recent collision events captured in the in-memory ring buffer (cap 50 events; older events live in `<sessionDir>/collision-events.jsonl`). Columns: relative timestamp, detection point, request, candidates (with schema badges), chosen winner, strategy, elapsed ms. A ⚡ marker flags events where the chosen candidate diverged from the `first-match` counterfactual — a real strategy divergence worth investigating. When telemetry capture is off, an inline reminder tells the tester how to opt in.

## @config pen on - Turn on Surface Pen Click Handler

Usage: `@config pen on`

## @config pen off - Turn off Surface Pen Click Handler

Usage: `@config pen off`

## @config ports - Lists the ports assigned to agents.

Usage: `@config ports`

## @const new - Create a new construction store

Usage: `@const new [<file>]`

### Arguments:

- &lt;file&gt; - (optional) File name to be created in the session directory or path to the file to be created. (type: string)

## @const load - Load a construction store from disk

Usage: `@const load [<file>]`

### Arguments:

- &lt;file&gt; - (optional) Construction file in the session directory or path to file (type: string)

## @const save - Save construction store to disk

Usage: `@const save [<file>]`

### Arguments:

- &lt;file&gt; - (optional) Construction file in the session directory or path to file (type: string)

## @const auto on - Turn on construction auto save

Usage: `@const auto on`

## @const auto off - Turn off construction auto save

Usage: `@const auto off`

## @const off - Disable construction store

Usage: `@const off`

## @const info - Show current construction store info

Usage: `@const info`

## @const list - List constructions

Usage: `@const list [--id <number>] [-p|--part <string>] [-m|--match <string>] [-b|--builtin] [-a|--all] [-v|--verbose]`

### Flags:

- --verbose -v : Verbose only. Includes part index, and list all string in match set (default: false)
- --all -a : List all string in match set (default: false)
- --builtin -b : List the construction in the built-in cache (default: false)
- --match -m &lt;string&gt; : Filter to constructions that has the string in the match set
- --part -p &lt;string&gt; : Filter to constructions that has the string match in the part name
- --id &lt;number&gt; : Construction id to list

## @const import - Import constructions from test data

Usage: `@const import [-t|--extended] [<file>...]`

### Arguments:

- &lt;file&gt; - (optional) Path to the construction file to import from. Load host specified test files if not specified. (type: string)

### Flags:

- --extended -t : Load host specified extended test files if no file argument is specified (default: false)

## @const prune - Prune out of date construction from the cache

Usage: `@const prune`

## @const delete - Delete a construction by id

Usage: `@const delete <namespace> <id>`

### Arguments:

- &lt;namespace&gt; - namespace the construction in (type: string)
- &lt;id&gt; - construction id to delete (type: number)

## @const builtin on - Turn on construction built-in cache

Usage: `@const builtin on`

## @const builtin off - Turn off construction built-in cache

Usage: `@const builtin off`

## @const merge on - Turn on construction merge

Usage: `@const merge on`

## @const merge off - Turn off construction merge

Usage: `@const merge off`

## @const wildcard on - Turn on wildcard matching

Usage: `@const wildcard on`

## @const wildcard off - Turn off wildcard matching

Usage: `@const wildcard off`

## @conversation new - Create a new conversation, optionally with a name

Usage: `@conversation new [<name>]`

### Arguments:

- &lt;name&gt; - (optional) Name for the new conversation (type: string)

## @conversation list - List all conversations

Usage: `@conversation list`

## @conversation info - Show info about the current conversation

Usage: `@conversation info`

## @conversation switch - Switch to a conversation by name (defaults to the next conversation in the list)

Usage: `@conversation switch [<name>]`

### Arguments:

- &lt;name&gt; - (optional) Name of the conversation to switch to. Omit to cycle to the next conversation. (type: string)

## @conversation prev - Switch to the previous conversation in the list (wraps around)

Usage: `@conversation prev`

## @conversation next - Switch to the next conversation in the list (wraps around)

Usage: `@conversation next`

## @conversation rename - Rename a conversation

Usage: `@conversation rename <nameOrNewName> [<newName>]`

### Arguments:

- &lt;nameOrNewName&gt; - With one argument, the new name for the current conversation; with two, the existing name of the conversation to rename. (type: string)
- &lt;newName&gt; - (optional) The new name when renaming a non-current conversation. (type: string)

## @conversation delete - Delete a conversation by name

Usage: `@conversation delete <name>`

### Arguments:

- &lt;name&gt; - Name of the conversation to delete (type: string)

## @debug - Start node inspector

Usage: `@debug`

## @dispatcher reason - Reason about a request

Usage: `@dispatcher reason [--engine <engine>] <request>`

### Arguments:

- &lt;request&gt; - Request to reason about (type: string)

### Flags:

- --engine &lt;engine&gt; : Reasoning engine to use: claude, copilot, or none. If omitted, uses the configured default from `@config execution reasoning`.

## @dispatcher reasoning - Reason about a request

Usage: `@dispatcher reasoning [--engine <engine>] <request>`  
Alias for `@dispatcher reason`. See above.

## @display - Send text to display

Usage: `@display [--inline] [--type <string>] [--speak] <text>...`

### Arguments:

- &lt;text&gt; - text to display (type: string)

## @dispatcher request - Translate and explain a request

Usage: `@dispatcher request [<request>]`

### Arguments:

- &lt;request&gt; - (optional) Request to translate (type: string)

## @dispatcher translate - Translate a request

Usage: `@dispatcher translate <request>`

### Arguments:

- &lt;request&gt; - Request to translate (type: string)

## @dispatcher explain - Explain a translated request with action

Usage: `@dispatcher explain [--concurrency <number>] [--filterReference] [--filterValueInRequest] [--repeat <number>] <requestAction>`

### Arguments:

- &lt;requestAction&gt; - Request to explain (type: string)

### Flags:

- --repeat <number> : Number of times to repeat the explanation (default: 1)
- --filterValueInRequest : Filter reference value for the explanation (default: false)
- --filterReference : Filter reference words (default: false)
- --concurrency <number> : Number of concurrent requests (default: 5)

## @display - Send text to display

Usage: `@display [--inline] [--type <string>] [--speak] <text>...`

### Arguments:

- &lt;text&gt; - text to display (type: string)

## @exit - Exit the program

Usage: `@exit`

### Flags:

- --speak : Speak the display for the host that supports TTS (default: false)
- --type <string> : Display type (default: text)
- --inline : Display inline (default: false)

## @email login - Log into the MS Graph to access email

Usage: `@email login`

## @email logout - Log out of MS Graph to access email

Usage: `@email logout`

## @grammar list - List grammar rules learned at runtime (optionally filtered by agent)

Usage: `@grammar list [<agent>]`

### Arguments:

- &lt;agent&gt; - (optional) Agent name to filter by, e.g. `list`, `player` (type: string)

Default subcommand of `@grammar` — `@grammar` alone is equivalent to `@grammar list`. Output is an HTML table with risk icons (munch / completion). Operates on the `persistedGrammarStore` (rules learned at runtime), not the static `.agr` files agents ship with.

## @grammar show - Show a stored grammar rule by ID

Usage: `@grammar show <id>`

### Arguments:

- &lt;id&gt; - Numeric ID of the rule to inspect (type: number)

Displays the rule's pattern (syntax-highlighted), anchor words, and risk analysis. IDs come from `@grammar list`.

## @grammar delete - Delete a stored grammar rule by ID

Usage: `@grammar delete <id>`

### Arguments:

- &lt;id&gt; - Numeric ID of the rule to delete (type: number)

Removes the rule from the persisted store and rebuilds the in-memory NFA for the affected schema so subsequent matches reflect the deletion immediately.

## @grammar clear - Clear stored grammar rules (optionally for a specific agent)

Usage: `@grammar clear [<agent>]`

### Arguments:

- &lt;agent&gt; - (optional) Agent name to clear rules for. Omit to clear all stored rules. (type: string)

## @grammar collisions - Scan all loaded agent grammars for cross-agent collisions

Usage: `@grammar collisions [--json <path>]`

### Flags:

- `--json <path>` - (optional) Write the structured scan result to this path as JSON, in addition to rendering the report.

Compiles every loaded schema's grammar to an NFA and runs a **product-construction intersection** across every cross-agent pair to find concrete inputs that both grammars accept. Each collision card shows:

- The two colliding schemas (badged with consistent colors)
- The **witness** — a concrete token sequence both grammars accept
- The colliding rule on each side (syntax-highlighted: literals green, wildcards blue/red, phrase sets teal, nested rules indigo)
- A **`Matches as`** line per side — runs the witness through each grammar's AST matcher and shows the resulting action object (`actionName({param: "value", …})`). Solves the case where a top-level rule renders as an opaque `<rules>` placeholder: the action interpretation pinpoints which inner alternative actually fired.

Witnesses containing synthetic `<TypeName>` placeholders (because at least one wildcard requires a custom entity type whose accepted strings can't be enumerated by the scanner) are flagged red and require manual review.

Schemas whose grammars don't compile are listed in a collapsible "Why N schema(s) were skipped" breakdown with per-reason counts (no grammar / wrong format / parse error / compile error) and a sample of schema names. Grammars that carry optimizer `tailCall` markers (the NFA compiler refuses these — they're an AST-only optimization) are auto-stripped before NFA compile so they participate in the scan, with the count surfaced in the breakdown.

The `--json <path>` flag writes a `CollisionScanResult` to disk for offline post-processing — same engine and shape as the standalone [`analyze-grammar-collisions` CLI](../../../ts/packages/actionGrammar/README.md#cli-analyze-grammar-collisions). The JSON is keyed by canonical `"schemaA|schemaB"` (alphabetical) so it's stable across runs and easy to diff in CI.

This is the proactive companion to runtime collision detection — see [Action Collision Detection](../../../ts/packages/dispatcher/dispatcher/README.md#action-collision-detection) in the dispatcher README.

## @feedback list - List recent user-feedback entries (most recent first)

Usage: `@feedback list [--limit <number>] [--all]`

### Flags:

- --limit &lt;number&gt; : Maximum number of entries to show (default: 20)
- --all : Include every entry; otherwise only the latest rating per request is shown (default: false)

## @feedback top - Aggregate user feedback — counts by rating and category

Usage: `@feedback top [--limit <number>]`

### Flags:

- --limit &lt;number&gt; : Top-N depth for the per-category breakdown (default: 10)

## @feedback filter - Filter feedback by rating, category, and/or date range

Usage: `@feedback filter [--rating <string>] [--category <string>] [--since <string>] [--until <string>] [--limit <number>] [--all]`

### Flags:

- --rating &lt;string&gt; : up | down | cleared
- --category &lt;string&gt; : wrong-agent | didnt-understand | bad-response | other
- --since &lt;string&gt; : ISO date (YYYY-MM-DD) — entries on/after this date
- --until &lt;string&gt; : ISO date (YYYY-MM-DD) — entries on/before this date
- --limit &lt;number&gt; : Maximum number of entries to show (default: 50)
- --all : Include every entry; otherwise only the latest rating per request (default: false)

## @feedback export - Export user-feedback entries to a local file (JSON or JSONL)

Usage: `@feedback export <file> [--format <string>] [--all]`

### Arguments:

- &lt;file&gt; - Destination path. Extension picks the format if `--format` is omitted: `.jsonl` → JSONL, anything else → JSON. (type: string)

### Flags:

- --format &lt;string&gt; : `json` | `jsonl` (overrides the path extension)
- --all : Include every entry; otherwise only the latest rating per request (default: false)

## @feedback count - Show the total number of feedback entries

Usage: `@feedback count`

## @history list - List history

Usage: `@history list`

## @history clear - Clear the history

Usage: `@history clear`

## @history delete - Delete a specific message from the chat history

Usage: `@history delete <index>`

### Arguments:

- &lt;index&gt; - Chat history index to delete. (type: number)

## @history entities list - Shows all of the entities currently in 'working memory.'

Usage: `@history entities list`

## @history entities delete - Delete entities from the chat history (working memory).

Usage: `@history entities delete <entityId>`
Arguments:

- &lt;entityId&gt; - The UniqueId of the entity (type: string)

## @history insert - Insert messages to chat history

Usage: `@history insert <messages>`

### Arguments:

- &lt;messages&gt; - Chat history messages to insert (type: json)

## @trace - Enable or disable trace namespaces

Usage: `@trace [-*|--clear] [<namespaces>...]`

### Arguments:

- &lt;namespaces&gt; - (optional) Namespaces to enable (type: string)

### Flags:

- --clear -\* : Clear all trace namespaces (default: false)

## @help - Show help

Usage: `@help [-a|--all] [<command>]`

### Arguments:

- &lt;command&gt; - (optional) command to get help for (type: string)

### Flags:

- --all -a : shows all commands (default: false)

## @random online - Uses the LLM to generate random requests.

Usage: `@random online`

## @random offline - Issues a random request from a dataset of pre-generated requests.

Usage: `@random offline`

## @notify info - Shows the number of notifications available

Usage: `@notify info`

## @notify clear - Clears notifications

Usage: `@notify clear`

## @notify show unread - Shows unread notifications

Usage: `@notify show unread`

## @notify show all - Shows all notifications

Usage: `@notify show all`

## @token summary - Get overall LLM usage statistics.

Usage: `@token summary`

## @token details - Gets detailed LLM usage statistics.

Usage: `@token details`

## @env all - Echos environment variables to the user interface.

Usage: `@env all`

## @env get - Echos the value of a named environment variable to the user interface

Usage: `@env get <name>`

### Arguments:

- &lt;name&gt; - The name of the environment variable. (type: string)

## @index list - List indexes

Usage: `@index list`

## @index create - Create a new index

Usage: `@index create <type> <name> <location>`
Arguments: - &lt;type&gt; - The type of index to create [image, email] (type: string) - &lt;name&gt; - Name of the index (type: string) - &lt;location&gt; - Location of the index (type: string)

## @index delete - Delete an index

Usage: `@index delete <name>`
Arguments:

- &lt;name&gt; - Name of the index to delete (type: string)

## @index info - Show index details

Usage: `@index info <name>`
Arguments:

- &lt;name&gt; - Name of the index (type: string)

## @install - Install an agent

Usage: `@install <name> <agent>`

### Arguments:

- &lt;name&gt; - Name of the agent (type: string)
- &lt;agent&gt; - Path of agent package directory or tar file to install (type: string)

## @mcpfilesystem server - Set the server arguments

Usage: `@mcpfilesystem server <allowedDirectories>...`

### Arguments:

- &lt;allowedDirectories&gt; - Allowed directories for the file system agent to access (type: string)

## @oracle request - Send a request to the Oracle

Usage: `@oracle request <question>`

### Arguments:

- &lt;question&gt; - Request for Oracle (type: string)

## @player spotify load - Load spotify user data

Usage: `@player spotify load <file>`

### Arguments:

- &lt;file&gt; - File to load (type: string)

## @player spotify login - Login to Spotify

Usage: `@player spotify login`

## @player spotify logout - Logout from Spotify

Usage: `@player spotify logout`

## @run - Run a command script file

Usage: `@run <input>`

### Arguments:

- &lt;input&gt; - command script file path (type: string)

## @session new - Create a new empty session

Usage: `@session new [--persist] [--keep]`

### Flags:

- --keep: Copy the current session settings in the new session (default: false)
- --persist: Persist the new session. Default to whether the current session is persisted.

## @session open - Open an existing session

Usage: `@session open <session>`

### Arguments:

- &lt;session&gt; - Name of the session to open. (type: string)

## @session reset - Reset config on session and keep the data

Usage: `@session reset`

## @session clear - Delete all data on the current sessions, keeping current settings

Usage: `@session clear`

## @session list - List all sessions. The current session is marked green.

Usage: `@session list`

## @session delete - Delete a session. If no session is specified, delete the current session and start a new session.

Usage: `@session delete [-a|--all] [<session>]`

### Arguments:

- &lt;session&gt; - (optional) Session name to delete (type: string)

### Flags:

- --all, -a: Delete all sessions

## @session info - Show info about the current session

Usage: `@session info`

## @shell show settings - Show shell settings

Usage: `@shell show settings`

## @shell show help - Show shell help

Usage: `@shell show help`

## @shell show metrics - Show shell metrics

Usage: `@shell show metrics`

## @shell show raw - Shows raw JSON shell settings

Usage: `@shell show raw`

## @shell set - Sets a specific setting with the supplied value

Usage: `@shell set <name> <value>`

### Arguments:

- &lt;name&gt; - Name of the setting to set (type: string)
- &lt;value&gt; - The new value for the setting (type: string)

## @shell run - Run Demo

Usage: `@shell run`

## @shell run interactive - Run Demo Interactive

Usage: `@shell run interactive`

## @shell break - Abort the currently running demo

Usage: `@shell break`

## @shell topmost - Always keep the shell window on top of other windows

Usage: `@shell topmost`

## @shell open - Show a new Web Content view

Usage: `@shell open <site>`

### Arguments:

- &lt;site&gt; - Alias or URL for the site of the open. (type: string)

## @shell close - Close the new Web Content view

Usage: `@shell close`

## @shell localWhisper off - Turn off Local Whisper integration

Usage: `@shell localWhisper off`

## @shell localWhisper on - Turn on Local Whisper integration.

Usage: `@shell localWhisper on`

## @shell theme light - Set the theme to light

Usage: `@shell theme light`

## @shell theme dark - Set the theme to dark

Usage: `@shell theme dark`

## @shell trash restore - Restore agent messages from the trash (un-hide everything that was sent to the bin via the trash icon)

Usage: `@shell trash restore`

## @shell trash flush - Permanently delete every message currently in the trash. They stay hidden and can no longer be restored

Usage: `@shell trash flush`

## @uninstall - Uninstall an agent

Usage: `@uninstall <name>`

### Arguments:

- &lt;name&gt; - Name of the agent (type: string)