Package agent_executor.common

• Package agent_executor.common
  • class AgentExecutionInfo
    • func addMessage
    • func addMessages
    • func addRequest
    • func addStepMessage
    • prop chatRound
    • prop events
    • func init
    • func markCancellation
    • prop messages
    • func removeLastMessage
    • prop retrievalInfo
    • func setAnswer
  • Skill built-in tools
    • class SkillTool
    • class ListDirectoryTool
    • class FileReadTool
    • class GlobSearchTool
    • class GrepSearchTool
    • class ShellExecuteTool
    • struct SkillToolsBuilder

class AgentExecutionInfo

func addMessage

func addMessage(msg: Message): Unit

• Description: Adds a single message to the execution.
• Parameters:
  • msg: Message, The message to be added.

func addMessages

func addMessages(msgs: Array<Message>): Unit

• Description: Adds multiple messages to the execution.
• Parameters:
  • msgs: Array<Message>, The messages to be added.

func addRequest

func addRequest(request: AgentRequest): Unit

• Description: Adds a new request to the execution and processes it.
• Parameters:
  • request: AgentRequest, The request to be added.

func addStepMessage

func addStepMessage(msg: Message): Unit

• Description: Adds a step message to the execution.
• Parameters:
  • msg: Message, The step message to be added.

prop chatRound

prop chatRound: ChatRound

• Description: Gets the chat round information for this execution.

prop events

prop events: EventStream

• Description: Gets the event stream associated with this execution.

func init

init(agent: Agent, request: AgentRequest)

• Description: Initializes the AgentExecutionInfo with the given agent and request.
• Parameters:
  • agent: Agent, The agent associated with this execution.
  • request: AgentRequest, The request associated with this execution.

func markCancellation

func markCancellation(): Unit

• Description: Marks the execution as cancelled.

prop messages

prop messages: MessageList

• Description: Gets the list of messages associated with this execution.

func removeLastMessage

func removeLastMessage(): Unit

• Description: Removes the last message from the execution.

prop retrievalInfo

prop retrievalInfo: ArrayList<RetrievalInfo>

• Description: Gets the retrieval information associated with this execution.

func setAnswer

func setAnswer(answer: String): Unit

• Description: Sets the answer for the execution.
• Parameters:
  • answer: String, The answer to be set.

Skill built-in tools

These tools are auto-injected by the @agent macro only when the skillRoot attribute is set. skillRoot defaults to None; with the default, none of these tools are added to the agent and skillBuiltinTools is ignored. When skillRoot is set, SkillTool (the runSkill tool) is always added, and the five general-purpose tools below are added when skillBuiltinTools is left at its default of true. To use these tools without enabling the skill system you would need to instantiate them directly and add them via agent.toolManager.addTool(...). See the skill package for the underlying loader and the dsl doc for the @agent attributes that drive injection.

When a user-supplied tool has the same name as a built-in, the user tool wins (built-ins are added first, then overwritten by name). SkillToolsBuilder logs a status line at agent construction so silent overrides are visible.

class SkillTool

public class SkillTool <: NativeFuncTool {
    public init(skillManager: SkillManager)
}

• Tool name: runSkill.
• Description: Loads the body of a skill and returns it so the agent can follow the skill's instructions. Parameters: name: String — the name of the skill to load.

class ListDirectoryTool

public class ListDirectoryTool <: NativeFuncTool {
    public init()
}

• Tool name: listDirectory.
• Description: Lists the entries of an absolute directory path. Directories are suffixed with /. Relative paths are rejected.

class FileReadTool

public class FileReadTool <: NativeFuncTool {
    public init()
}

• Tool name: fileRead.
• Description: Reads a file at an absolute path and returns its contents wrapped in a <file-content> block. Optional startLine / endLine parameters narrow the result to a 1-based inclusive line range.

class GlobSearchTool

public class GlobSearchTool <: NativeFuncTool {
    public init()
}

• Tool name: globSearch.
• Description: Recursively finds files matching a glob pattern under an absolute path. Skips common build/VCS directories (.git, node_modules, target, build, dist, .cache).

class GrepSearchTool

public class GrepSearchTool <: NativeFuncTool {
    public init()
}

• Tool name: grepSearch.
• Description: Searches file contents for a regex pattern under an absolute path. Supports an optional fileType glob filter.

class ShellExecuteTool

public class ShellExecuteTool <: NativeFuncTool {
    public init()
}

• Tool name: shellExecute.
• Description: Runs a shell command (PowerShell on Windows, /bin/sh elsewhere). Optional workDir must be an absolute path. Output is truncated to ~30 KB per stream; default timeout is 180 s (max 600 s).

struct SkillToolsBuilder

public struct SkillToolsBuilder {
    public static func build(skillManager: SkillManager,
                             builtinEnabled!: Bool): Array<Tool>
    public static func computeSkillToolStatus(tm: ToolManager,
                                              builtinEnabled: Bool): Array<String>
    public static func logSkillToolStatus(agentName: String,
                                          tm: ToolManager,
                                          builtinEnabled: Bool): Unit
}

• Description: Builds and reports on the skill built-in tool set.
• build returns [SkillTool] when builtinEnabled is false, otherwise SkillTool plus the five general-purpose tools above.
• computeSkillToolStatus returns the status messages that logSkillToolStatus writes at INFO level — useful in tests.