Built-in Tools

Silicon Worker ships with a set of built-in tools that let the AI assistant read and write files, operate the desktop and browser, look things up, and manage skills and subtasks. They're grouped by capability domain below, each listing what it does and its risk level.

Risk level decides whether your consent is requested first (see Tools & risk):

Level	Meaning
Safe	Read-only or no side effects; usually runs automatically.
Low	Has side effects but limited impact; asking depends on your permission mode.
High	May cause irreversible changes (overwriting files, running commands, deleting data); usually prompts for confirmation first.

Some tools are judged dynamically: the same tool is Safe to query, Low to create/update, and High to delete.

File operations

Tool	What it does	Risk
Read file	Read the contents of a text file	Safe
Write file	Create or overwrite a file	High
Edit file	Make targeted, patch-like edits to a file	High
Find files	Find files by glob / pattern	Safe
Search text	Search text inside files using regex	Safe

Command execution

Tool	What it does	Risk
Run command	Execute a shell command in the workspace	High

Desktop

Tool	What it does	Risk
Desktop control	Recognize UI elements and perform desktop automation: click, type, scroll, screenshot, wait	Safe to observe, Low and up to act

Desktop control needs system permissions — see System Permissions and Desktop & browser automation.

Browser automation

Tool	What it does	Risk
Browser control	Open pages, read page content, click, fill forms, select dropdowns, scroll, extract text, go back	Safe to Low per action
Switch tab	Switch the active browser tab	Safe
Close tab	Close a given browser tab	Safe

High-risk actions like submitting forms or clicking delete are detected automatically and confirmed first.

Web

Tool	What it does	Risk
Web search	Search the web for content	Safe
Fetch web page	Fetch and parse HTML / JSON from a page or API	Safe

Apple apps

Tool	What it does	Risk
Calendar	View / create / update / delete calendar events	Dynamic (view Safe, create/update Low, delete High)
Reminders	View / create / update / delete reminders	Dynamic
Notes	View / create / update / delete notes	Dynamic

Apple-app capabilities need the matching system permissions — see System Permissions.

Knowledge base

Tool	What it does	Risk
Search knowledge base	Full-text search across the bound knowledge bases	Safe

Creating and binding knowledge bases is covered in Knowledge base.

Skills & experts

Tool	What it does	Risk
Install skill	Install a skill from a zip or directory	Low
Load skill	Reference and load a skill by name	Safe
Create expert	Create an expert (a reusable role) from a prompt	Low
Create team	Compose a team from multiple experts	Low

For the concepts behind these, see Skills and Experts & teams.

Delegation & collaboration

Tool	What it does	Risk
Dispatch sub-agent	Spawn a child assistant to handle a subtask in parallel (non-blocking)	Low
Collect results	Gather the results returned by child assistants	Safe

Assistant control

Tool	What it does	Risk
Ask you	Ask you one or more questions to gather information	Safe
Propose plan	Propose a multi-step plan for your approval	Safe
Propose personality update	Propose an update to a companion's personality (needs your approval)	Low

Tasks & artifacts

Tool	What it does	Risk
Update to-dos	Manage the current session's to-do list	Safe
Update tasks	Update the project task list	Safe
Add artifact	Create or attach result artifacts (files, images, Markdown)	Safe

Memory

Tool	What it does	Risk
Remember	Store a fact in long-term memory	Safe

How memory works is covered in Memory.

Plugins

Tool	What it does	Risk
Install plugin	Install a plugin (may include skills, experts, hooks, MCP servers)	Low

Tool discovery

Tool	What it does	Risk
Find tools	Activate more tools on demand by keyword or name	Safe

Extending with more tools

Beyond the built-in tools, Silicon Worker can connect external tools via MCP, exposing third-party services as callable tools. See Connect MCP.