ScienceOL Docs
  • English

      • Tools

      Tools are capabilities that agents can invoke during execution. SciLaxy ships with 40+ builtin tools across several categories.

      Tools are enabled per-agent in the agent configuration. When an agent has a tool enabled, the LLM can choose to call it during the tool-calling loop.

      Info

      Enabling a tool on an Agent does not grant an execution environment. A conversation with no Project has no command, project-filesystem, or sandbox execution tools. Bind a Project with an available execution target before asking the Agent to run code or operate on project files.

      Search & Research

      ToolDescription
      Web SearchSearch the web and return relevant results
      Academic ResearchSearch academic databases for papers and citations
      Literature SearchFind and analyze scientific literature
      URL FetchFetch and extract content from any URL

      Code Execution

      ToolDescription
      Sandbox ExecuteRun code in an isolated sandbox environment
      Sandbox File WriteWrite files to the sandbox filesystem
      Sandbox File ReadRead files from the sandbox filesystem
      Sandbox Directory ListList files and directories in the sandbox

      The sandbox tools execute code in isolated environments. See Sandbox for details on the execution environment.

      These tools are exposed only for a Project-bound conversation whose execution target is available. SciLaxy does not create or fall back to a cloud sandbox for a conversation with no Project.

      Knowledge & Memory

      ToolDescription
      Knowledge ReadRetrieve documents from the agent's linked knowledge base
      Knowledge WriteAdd or update documents in the knowledge base
      Memory ReadRead from the agent's core memory sections
      Memory WriteUpdate core memory (user summary, preferences, context, rules)
      Memory SearchSearch the agent's long-term memory store

      See Knowledge Bases and Memory for more details.

      File Operations

      ToolDescription
      File ReadRead uploaded files (PDF, documents, images, etc.)
      File UploadHandle user-uploaded files

      The file reader supports multiple formats including PDF, DOCX, plain text, and images.

      Reading an explicitly uploaded attachment is separate from project-filesystem access. Without a Project, an Agent cannot list, read, or write a device, Project, or sandbox filesystem.

      Multimedia

      ToolDescription
      Image GenerationGenerate images using AI models
      Video GenerationGenerate videos using Google Veo via Vertex AI

      Agent Management

      ToolDescription
      Create AgentDynamically create new agents during execution
      Delegate to Sub-AgentDelegate a subtask to another agent
      Scheduled TaskCreate scheduled tasks for recurring automation

      User Interaction

      ToolDescription
      Ask User QuestionPause execution and ask the user for input

      This tool allows agents to request clarification or additional information from the user mid-execution, then continue with the response.

      Enabling Tools

      Tools are configured per-agent in the agent settings. You can enable or disable individual tools based on the agent's purpose:

      1. Open the agent configuration
      2. Navigate to the Tools section
      3. Toggle the tools you want the agent to have access to
      Info

      SciLaxy shows the capabilities available to the current workspace. Connection authorization is completed in the product interface; Project execution tools also require their selected execution target to be online.