@isdk/ai-tool 0.8.2 → 0.9.0

package/docs/api/interfaces/TextifyOptions.md

@@ -0,0 +1,108 @@
+[**@isdk/ai-tool**](../README.md)
+
+***
+
+[@isdk/ai-tool](../globals.md) / TextifyOptions
+
+# Interface: TextifyOptions
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:53](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L53)
+
+文本化配置选项
+
+## Description
+
+该工具旨在生成易于人类阅读的、类 YAML 格式的文本表示。
+
+**核心渲染规则：**
+1. **对象数组 (Object Array)**: 采用 YAML 风格的紧凑模式。
+   当对象作为数组项时，会忽略对象自身的 `objectPrefix`，并将第一个键值对紧跟在数组前缀 `-` 后。
+   示例：`[{a: 1, b: 2}]` -> `- a: 1\n  b: 2`
+
+2. **嵌套数组 (Nested Array)**: 采用阶梯式 (Block Style)。
+   内层数组会从新的一行开始，并相对于父级缩进。
+   示例：`[[1, 2]]` -> `-\n  - 1\n  - 2`
+
+3. **多行字符串对齐**: 自动计算前缀和键名的长度，确保多行文本的后续行与首行左对齐。
+
+**高级特性：**
+- **智能行内化 (Inline Style)**: 自动将短对象或数组转换为 `{a: 1}` 格式。
+- **混合模式 (Mixed Mode)**: 根据内容长度自动决定嵌套项是行内还是块状显示。
+- **自动引用 (Auto Quoting)**: 智能识别需要引号包裹的字符串（包含特殊字符、保留字或数字外观）。
+
+## Properties
+
+### arrayPrefix?
+
+> `optional` **arrayPrefix**: `string`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:57](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L57)
+
+数组项前缀 (默认：'-')
+
+***
+
+### ensureNewLineForMultiline?
+
+> `optional` **ensureNewLineForMultiline**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:66](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L66)
+
+是否确保多行字符串强制换行
+(默认：false)
+
+***
+
+### indent?
+
+> `optional` **indent**: `string` \| `number`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:55](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L55)
+
+缩进单位，可以是空格数或具体的字符串 (默认：2 个空格)
+
+***
+
+### inlineStyle?
+
+> `optional` **inlineStyle**: [`TextifyInlineStyleMode`](../type-aliases/TextifyInlineStyleMode.md) \| [`TextifyInlineOptions`](TextifyInlineOptions.md)
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:73](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L73)
+
+行内化配置
+- 'auto': 基于阈值自动切换 (默认)
+- 'always': 强制行内化 (遇到换行符除外)
+- 'never': 强制块状化
+
+***
+
+### maxDepth?
+
+> `optional` **maxDepth**: `number`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:61](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L61)
+
+最大递归深度，防止过深 (默认：10)
+
+***
+
+### objectPrefix?
+
+> `optional` **objectPrefix**: `string`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:59](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L59)
+
+对象键值对前缀 (默认：'*')
+
+***
+
+### stringQuoting?
+
+> `optional` **stringQuoting**: [`TextifyStringQuoting`](../type-aliases/TextifyStringQuoting.md)
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/textify.ts:80](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/textify.ts#L80)
+
+字符串引用策略 (是否加双引号)
+- 'auto': 遇到特殊字符 (:, #, [, ], {, }, *, -, ")、首尾空格、保留字或数字外观时自动加引号 (默认)
+- 'always': 所有字符串始终加引号
+- 'never': 除非结构必须，否则不加引号

package/docs/api/interfaces/ToolFuncContext.md

@@ -0,0 +1,73 @@
+[**@isdk/ai-tool**](../README.md)
+
+***
+
+[@isdk/ai-tool](../globals.md) / ToolFuncContext
+
+# Interface: ToolFuncContext
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:113
+
+Execution context for a tool function.
+
+## Indexable
+
+\[`key`: `string`\]: `any`
+
+Allows users to extend arbitrary properties.
+
+## Properties
+
+### binding?
+
+> `optional` **binding**: `"auto"` \| `"early"` \| `"late"`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:125
+
+The binding strategy for internal dependencies (runAsSync).
+- 'early': Always use pre-bound instances from 'depends'.
+- 'late': Always resolve from rootRegistry (forced polymorphism).
+- 'auto': Use 'late' if rootRegistry shadows the dependency, else 'early' (Safe Default).
+
+***
+
+### inheritContext?
+
+> `optional` **inheritContext**: `boolean`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:135
+
+Whether to allow context inheritance/propagation in nested calls.
+Defaults to true.
+
+***
+
+### isolated?
+
+> `optional` **isolated**: `boolean`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:130
+
+Whether to enable independent execution scope.
+If true, a temporary instance will be created via Object.create(this) to isolate concurrency.
+
+***
+
+### rootRegistry?
+
+> `optional` **rootRegistry**: *typeof* [`ToolFunc`](../classes/ToolFunc.md)
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:118
+
+The entry-point registry class that initiated the call chain.
+Used for late-binding dependency resolution in hierarchical registries.
+
+***
+
+### signal?
+
+> `optional` **signal**: `AbortSignal`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:139
+
+Standard Web AbortSignal for propagating cancellation signals.

package/docs/api/interfaces/ToolFuncPackage.md

@@ -6,7 +6,7 @@
 
 # Interface: ToolFuncPackage
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/tool-func.ts:216](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/tool-func.ts#L216)
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:368
 
 Describes a package of tool functions, including methods for registration and unregistration.
 
@@ -16,7 +16,7 @@ Describes a package of tool functions, including methods for registration and un
 
 > **name**: `string`
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/tool-func.ts:221](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/tool-func.ts#L221)
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:373
 
 The name of the tool function package.
 
@@ -26,7 +26,7 @@ The name of the tool function package.
 
 > **register**: (`data?`) => `void`
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/tool-func.ts:226](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/tool-func.ts#L226)
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:378
 
 A method to register all functions within the package.
 
@@ -48,7 +48,7 @@ Optional data to pass to the registration process.
 
 > `optional` **unregister**: () => `void`
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/tool-func.ts:230](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/tool-func.ts#L230)
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:382
 
 An optional method to unregister all functions within the package.
 

package/docs/api/interfaces/ToolFuncRegistryIsolateOptions.md

@@ -0,0 +1,41 @@
+[**@isdk/ai-tool**](../README.md)
+
+***
+
+[@isdk/ai-tool](../globals.md) / ToolFuncRegistryIsolateOptions
+
+# Interface: ToolFuncRegistryIsolateOptions
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:338
+
+Options for isolating a ToolFunc registry.
+
+## Properties
+
+### aliases?
+
+> `optional` **aliases**: `boolean`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:342
+
+Whether to isolate the alias map (default: true).
+
+***
+
+### items?
+
+> `optional` **items**: `boolean`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:340
+
+Whether to isolate the main function registry (default: true).
+
+***
+
+### refCounts?
+
+> `optional` **refCounts**: `boolean`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:344
+
+Whether to isolate the reference counts (default: true).

package/docs/api/interfaces/UnregisterOptions.md

@@ -0,0 +1,56 @@
+[**@isdk/ai-tool**](../README.md)
+
+***
+
+[@isdk/ai-tool](../globals.md) / UnregisterOptions
+
+# Interface: UnregisterOptions
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:387
+
+Options for unregistering a tool function.
+
+## Properties
+
+### decrement?
+
+> `optional` **decrement**: `"once"` \| `"all"`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:400
+
+How to handle the reference count.
+- 'once' (default): Decrement the count by one.
+- 'all': Completely remove the reference count entry.
+
+#### Default
+
+```ts
+force ? 'all' : 'once'
+```
+
+***
+
+### force?
+
+> `optional` **force**: `boolean`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:392
+
+If true, force physical removal from the registry even if references exist.
+Also defaults the `decrement` option to `'all'` if not specified.
+
+***
+
+### scope?
+
+> `optional` **scope**: `"all"` \| `"local"` \| `"inherited"`
+
+Defined in: @isdk/ai-tools/packages/tool-func/dist/index.d.ts:410
+
+The scope of unregistration in a hierarchical registry:
+- 'local' (default): Only remove if the item is "owned" by the current scope.
+  Ownership is defined by having an 'own' property in items, aliases, OR reference counts.
+  Note: Including reference counts ensures that circular dependencies are correctly cleaned up
+  even after the primary instance is removed from the items list during an override.
+- 'inherited': Search up the prototype chain and remove the first match found.
+- 'all': Remove all occurrences found in the entire prototype chain.

package/docs/api/type-aliases/AIChatContentPart.md

@@ -6,6 +6,6 @@
 
 # Type Alias: AIChatContentPart
 
-> **AIChatContentPart** = [`AIChatContentPartText`](../interfaces/AIChatContentPartText.md) \| [`AIChatContentPartImage`](../interfaces/AIChatContentPartImage.md)
+> **AIChatContentPart** = [`AIChatContentPartBase`](../interfaces/AIChatContentPartBase.md) \| [`AIChatContentPartText`](../interfaces/AIChatContentPartText.md) \| [`AIChatContentPartImage`](../interfaces/AIChatContentPartImage.md)
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:83](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/chat.ts#L83)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:93](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L93)

package/docs/api/type-aliases/AIChatMessageParam.md

@@ -8,4 +8,4 @@
 
 > **AIChatMessageParam** = [`AIChatMessageParamBase`](../interfaces/AIChatMessageParamBase.md) \| [`AIChatSystemMessageParam`](../interfaces/AIChatSystemMessageParam.md) \| [`AIChatUserMessageParam`](../interfaces/AIChatUserMessageParam.md) \| [`AIChatAssistantMessageParam`](../interfaces/AIChatAssistantMessageParam.md) \| [`AIChatToolMessageParam`](../interfaces/AIChatToolMessageParam.md)
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:47](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/chat.ts#L47)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:57](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L57)

package/docs/api/type-aliases/AIChatRole.md

@@ -6,6 +6,6 @@
 
 # Type Alias: AIChatRole
 
-> **AIChatRole** = *typeof* [`AIChatRoles`](../variables/AIChatRoles.md)\[`number`\] & `string`
+> **AIChatRole** = *typeof* [`AIChatRoles`](../variables/AIChatRoles.md)\[`number`\] \| `string` & `object`
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:45](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/chat.ts#L45)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:44](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L44)

package/docs/api/type-aliases/AIChatTemplateCaps.md

@@ -0,0 +1,239 @@
+[**@isdk/ai-tool**](../README.md)
+
+***
+
+[@isdk/ai-tool](../globals.md) / AIChatTemplateCaps
+
+# Type Alias: AIChatTemplateCaps
+
+> **AIChatTemplateCaps** = `object`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:202](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L202)
+
+Capabilities detected for a chat template (ChatML-like) with respect to
+OpenAI-compatible function/tool calling in llama.cpp-style pipelines.
+
+Each flag describes what the template can (or must) do when rendering messages/tools,
+so your caller can adapt message shaping, tool-calling strategy, and fallbacks.
+
+Detection is typically performed by rendering small “probe” conversations against
+the template and inspecting the resulting text.
+
+## Properties
+
+### requiresNonNullContent
+
+> **requiresNonNullContent**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:327](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L327)
+
+Whether the template forbids `assistant.content` being `null` (requires an empty string instead).
+
+#### Remarks
+
+- true: Use `content: ""` as a placeholder when the assistant turn only carries `tool_calls`.
+- false: `null` or `""` both work.
+
+#### Impact
+
+- Prevents template branches from swallowing messages when content is `null`.
+
+***
+
+### requiresObjectArguments
+
+> **requiresObjectArguments**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:276](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L276)
+
+Whether the template expects `function.arguments` to be a JSON object (not a JSON-string literal).
+
+#### Remarks
+
+- true: Pass arguments as an object: `{ arguments: { ... } }`. Supplying a stringified JSON may double-escape or break parsing.
+- false: Both object and string are accepted (object is still recommended for stricter grammar alignment).
+
+#### Impact
+
+- Honor this when constructing assistant messages that contain `tool_calls`.
+
+#### Example
+
+```ts
+const args = caps.requiresObjectArguments ? { city: "SF" } : JSON.stringify({ city: "SF" });
+```
+
+***
+
+### requiresTypedContent
+
+> **requiresTypedContent**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:220](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L220)
+
+Whether the template requires OpenAI-style "typed content" parts instead of a plain string.
+
+#### Remarks
+
+- true: A message like `{ content: "..." }` is ignored, while
+  `{ content: [{ type: "text", text: "..." }] }` renders correctly.
+- false: Both plain strings and typed parts render.
+
+#### Impact
+
+- If true, wrap all string content as text parts before rendering:
+  `{ content: [{ type: "text", text }] }`.
+
+#### Example
+
+```ts
+if (caps.requiresTypedContent && typeof msg.content === "string") {
+  msg.content = [{ type: "text", text: msg.content }];
+}
+```
+
+***
+
+### supportsParallelToolCalls
+
+> **supportsParallelToolCalls**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:291](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L291)
+
+Whether a single assistant message can contain multiple tool calls and be rendered correctly.
+
+#### Remarks
+
+- true: The template can serialize an array of calls in one assistant turn.
+- Note this only opens the possibility; the runtime must also allow it
+  (e.g., request `parallel_tool_calls: true` and use a grammar with an array top rule),
+  and the model must actually choose to emit multiple calls.
+- false: Even if the API flag is set, only 0/1 call will practically work per turn.
+
+#### Impact
+
+- Enable/disable “same-turn multi-call” orchestration accordingly.
+
+***
+
+### supportsSystemRole
+
+> **supportsSystemRole**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:234](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L234)
+
+Whether the template actually includes role=system messages in the rendered prompt.
+
+#### Remarks
+
+- true: You can pass `messages: [{ role: "system", ... }, ...]` and expect
+  the system text to appear at the top of the prompt.
+- false: The template ignores system; move your “system rules” into the first user/assistant turn or edit the template.
+
+#### Impact
+
+- If false, critical behavior instructions should be injected via the template itself
+  or merged into the first turn.
+
+***
+
+### supportsToolCallId
+
+> **supportsToolCallId**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:315](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L315)
+
+Whether the template surfaces `tool_call_id` when rendering role=tool results, enabling strict mapping back to the call.
+
+#### Remarks
+
+- true: Each tool output can be visibly tied to its originating call (important for multi-call turns).
+- false: The mapping is implicit; include the call id in the text or restrict to single calls per turn.
+
+#### Impact
+
+- Strongly recommended when parallel/multiple tool calls are possible.
+
+***
+
+### supportsToolCalls
+
+> **supportsToolCalls**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:261](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L261)
+
+Whether the template properly renders an assistant message that contains `tool_calls` (OpenAI format).
+
+#### Remarks
+
+- true: An assistant message with `tool_calls: [{ type: "function", function: { name, arguments }}, ...]`
+  gets serialized in the template’s tool-call section.
+- false: The template does not serialize tool calls correctly; switch template/handler,
+  fall back to a Generic handler, or use the “pure JSON intent” two-turn strategy.
+
+#### Impact
+
+- If false, do not rely on OpenAI `tool_calls` in a single assistant turn.
+
+***
+
+### supportsToolResponses
+
+> **supportsToolResponses**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:303](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L303)
+
+Whether the template renders role=tool messages (tool/function execution results) back into the prompt context.
+
+#### Remarks
+
+- true: The model can see tool outputs (often near the next assistant generation).
+- false: Tool results aren’t visible; inject them as user/system text instead, or update the template to render tool messages.
+
+#### Impact
+
+- Critical for the standard call→execute→return→answer loop to converge.
+
+***
+
+### supportsTools
+
+> **supportsTools**: `boolean`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:247](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L247)
+
+Whether the template consumes the top-level `tools` array (function catalog) and renders it into the prompt.
+
+#### Remarks
+
+- true: The model sees a readable list of available functions (names, descriptions, schemas).
+- false: The template doesn’t show the catalog; tool-calling can still work, but the model
+  may be less aware of function options unless you add guidance text.
+
+#### Impact
+
+- If false, consider adding a short “How to call tools” section or a generated example to the template.
+
+***
+
+### toolCallExample?
+
+> `optional` **toolCallExample**: `string`
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:342](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L342)
+
+A generated “how to call a tool” example snippet, inferred only when the template doesn’t render the tools catalog.
+
+#### Remarks
+
+- Present when ChatTemplateCaps.supportsTools is false and the probe could
+  derive a stable example by diffing “prefix only” vs “prefix + a fake tool call”.
+- Useful to inject into the template/system text to teach the expected call format.
+
+#### Example
+
+```ts
+if (!caps.supportsTools && caps.toolCallExample) {
+  system += "\n\nTool call example:\n" + caps.toolCallExample;
+}
+```

package/docs/api/type-aliases/AIChatToolChoiceParam.md

@@ -8,4 +8,4 @@
 
 > **AIChatToolChoiceParam** = `"auto"` \| `"none"` \| `"required"` \| [`AIChatToolChoiceObject`](../interfaces/AIChatToolChoiceObject.md)
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:142](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/chat.ts#L142)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:159](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L159)

package/docs/api/type-aliases/AIMessageType.md

@@ -8,4 +8,4 @@
 
 > **AIMessageType** = *typeof* [`AIMessageTypes`](../variables/AIMessageTypes.md)\[`number`\]
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:38](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/chat.ts#L38)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:38](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L38)

package/docs/api/type-aliases/AIModelNameRule.md

@@ -8,7 +8,7 @@
 
 > **AIModelNameRule** = `string` \| `RegExp` \| [`AIModelNameRuleFn`](AIModelNameRuleFn.md)
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/consts.ts:29](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/consts.ts#L29)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/consts.ts:18](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/consts.ts#L18)
 
 Defines a single rule for matching an AI model name. It can be a simple string,
 a regular expression, or a custom function.

package/docs/api/type-aliases/AIModelNameRuleFn.md

@@ -8,7 +8,7 @@
 
 > **AIModelNameRuleFn** = (`name`) => `string` \| `RegExpExecArray` \| `undefined`
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/consts.ts:23](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/consts.ts#L23)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/consts.ts:12](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/consts.ts#L12)
 
 Defines the signature for a function that processes an AI model's name,
 typically for matching or validation purposes.

package/docs/api/type-aliases/AIModelNameRules.md

@@ -8,6 +8,6 @@
 
 > **AIModelNameRules** = [`AIModelNameRule`](AIModelNameRule.md) \| [`AIModelNameRule`](AIModelNameRule.md)[]
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/consts.ts:33](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/consts.ts#L33)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/consts.ts:22](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/consts.ts#L22)
 
 Represents a collection of one or more AI model name matching rules.

package/docs/api/type-aliases/AITextGenerationFinishReason.md

@@ -8,4 +8,4 @@
 
 > **AITextGenerationFinishReason** = *typeof* [`AITextGenerationFinishReasons`](../variables/AITextGenerationFinishReasons.md)\[`number`\]
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:14](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/chat.ts#L14)
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/chat.ts:14](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/chat.ts#L14)

package/docs/api/type-aliases/ActionName.md

@@ -8,6 +8,6 @@
 
 > **ActionName** = *typeof* [`ActionNames`](../variables/ActionNames.md)\[`number`\]
 
-Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/consts.ts:15](https://github.com/isdk/ai-tool.js/blob/e883e341c67e937e7d3a3e95e8bc56844896f5a3/src/utils/consts.ts#L15)
+Defined in: @isdk/ai-tools/packages/tool-rpc/dist/index-359784IS.d.ts:14
 
 Represents a valid action name for a remote tool, derived from the `ActionNames` tuple.

package/docs/api/type-aliases/ArgProcessor.md

@@ -0,0 +1,24 @@
+[**@isdk/ai-tool**](../README.md)
+
+***
+
+[@isdk/ai-tool](../globals.md) / ArgProcessor
+
+# Type Alias: ArgProcessor()
+
+> **ArgProcessor** = (`ctx`) => `any` \| `Promise`\<`any`\>
+
+Defined in: [@isdk/ai-tools/packages/ai-tool/src/utils/command-parser/types.ts:142](https://github.com/isdk/ai-tool.js/blob/d10fb4cda65fc1975152a2c3ab327ecab008dea1/src/utils/command-parser/types.ts#L142)
+
+Type definition for an argument processor function.
+It can return a plain value (for standard evaluation) or a Protocol-compliant object.
+
+## Parameters
+
+### ctx
+
+[`CmdArgContext`](../interfaces/CmdArgContext.md)
+
+## Returns
+
+`any` \| `Promise`\<`any`\>