CoWork OS

342 tools. 211 can modify or destroy data without limits.

17 destructive tools with no built-in limits. Policy required.

Last updated:

211 can modify or destroy data
131 read-only
342 tools total

Community server · catalogue entry checked 07/07/2026

How to control CoWork OS ↓

What CoWork OS exposes to your agents

Read (131) Write / Execute (194) Destructive / Financial (17)
Critical Risk

The most dangerous CoWork OS tools

211 of CoWork OS's 342 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

How to control CoWork OS

PolicyLayer is an MCP gateway — it sits between your AI agents and CoWork OS, and nothing reaches the server without passing your rules. These are the rules we recommend:

Block financial tools by default
{
  "domain_register": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Requires human approval."
      }
    ]
  }
}

Financial tools should be explicitly enabled per use case, not open by default.

Deny destructive operations
{
  "cancel_agent": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations
{
  "acknowledge_mention": {
    "limits": [
      {
        "counter": "acknowledge_mention_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations
{
  "acp_discover": {
    "limits": [
      {
        "counter": "acp_discover_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

  1. Create a free account and register CoWork OS — nothing to install.
  2. Add these rules — paste them, or build them visually. Tune the limits to your setup.
  3. Point your MCP client (Claude, Cursor, anything) at your gateway URL.
ENFORCE POLICY ON COWORK OS →

Instant setup, no code required.

All 342 CoWork OS tools

DESTRUCTIVE 15 tools
Destructive cancel_agent Cancel a descendant child agent task. This tool only works for tasks spawned by the current task (descendants) Destructive cancel_video_generation_job Cancel an in-progress video generation job. Not all providers support cancellation; an error will be returned Destructive cloud_sandbox_delete Delete a cloud sandbox and free resources. Destructive delete_file Delete a file (requires user approval) Destructive disable_macos_launch_agents Unload and move matching user LaunchAgent plists into ~/Library/LaunchAgents.disabled-by-cowork. Destructive domain_dns_delete Delete a DNS record from a domain. Destructive google-workspace.calendar_event_delete Delete a Google Calendar event. Confirm with the user before calling. Destructive google-workspace.slides_delete_slide Delete a slide from a Google Slides presentation. Confirm with the user before calling. Destructive google-workspace.tasks_clear_completed Clear completed tasks from a Google Tasks task list. Confirm with the user before calling. Destructive google-workspace.tasks_delete Delete a Google Tasks task. Confirm with the user before calling. Destructive google-workspace.tasks_lists_delete Delete a Google Tasks task list. Confirm with the user before calling. Destructive kg_delete_edge Delete a relationship (edge) from the knowledge graph. Destructive kg_delete_entity Delete an entity and all its relationships and observations from the knowledge graph. Destructive skill_delete Delete a skill. Only managed and workspace skills can be deleted (not bundled). Destructive supermemory_forget Forget a Supermemory entry by ID or exact content. Use this when an external memory is outdated or wrong.
EXECUTE 74 tools
Execute blender.render_view Render a camera view to an image Execute browser_act_batch Execute a batch of browser actions in sequence. Use for multi-step interactions (e.g. fill form, click submit, Execute browser_attach Attach to an existing Chrome browser session via Chrome DevTools Protocol. Execute browser_back Go back in browser history Execute browser_click Click on an element on the page Execute browser_close Close the browser Execute browser_close_tab Close a browser tab by tab id Execute browser_drag Drag from one Browser V2 snapshot ref to another Execute browser_emulate Set the visible in-app browser viewport/device emulation for responsive testing. Execute browser_evaluate Execute JavaScript code in the browser context Execute browser_fill Fill a form field with text Execute browser_forward Go forward in browser history Execute browser_handle_dialog Accept or dismiss the latest JavaScript dialog in the browser Execute browser_hover Move the browser pointer over an element, preferably by browser_snapshot ref Execute browser_navigate Navigate the browser to a URL. By default, CoWork uses background/headless browser control unless settings or Execute browser_press Press a keyboard key (e.g., Enter, Tab, Escape) Execute browser_reload Reload the current page Execute browser_scroll Scroll the page Execute browser_select Select an option from a dropdown Execute browser_switch_tab Switch to a browser tab by tab id Execute browser_trace_start Start a lightweight browser trace for diagnostics Execute browser_trace_stop Stop the active lightweight browser trace Execute browser_type Type text character by character (useful for search boxes with autocomplete) Execute browser_wait Wait for an element to appear on the page Execute canvas_eval Execute JavaScript code in the canvas context. Execute canvas_open_url Open a remote web page inside the canvas window for full in-app browsing. Execute canvas_show OPTIONAL: Open the canvas in a separate interactive window. Execute click Click inside the current controlled window using screenshot-relative coordinates. Execute cloud_sandbox_create Create a new cloud sandbox (Linux VM). Returns sandbox ID and connection info. Execute cloud_sandbox_exec Execute a shell command in a cloud sandbox. Returns stdout, stderr, and exit code. Execute comfyui.submit_flux_photoreal_pass Submit a Flux-style photoreal workflow graph with project metadata Execute comfyui.submit_workflow Submit a ComfyUI workflow JSON graph Execute compile_latex Compile a workspace .tex file into a PDF using a system LaTeX engine. Execute double_click Double-click inside the current controlled window using screenshot-relative coordinates. Returns a fresh scree Execute drag Drag through a series of screenshot-relative points inside the current controlled window. Returns a fresh scre Execute execute_code Execute Python, JavaScript, or shell code in a sandboxed process with a configurable timeout. Execute generate_image Generate an image from a text description using AI. CoWork OS will pick the best configured provider by defaul Execute generate_narration_audio Generate narrated MP3 audio from text using the configured voice service. Execute generate_video Generate a video from a text prompt (or an image/video reference) using AI. CoWork OS will pick the best confi Execute git_merge_to_base Merge the current worktree branch back to the base branch. Only available in worktree isolation mode. Use when Execute google-workspace.slides_batch_update Run raw Google Slides presentations.batchUpdate requests for advanced edits Execute http_request Make HTTP requests like curl. Supports all HTTP methods, custom headers, and request bodies. Execute keypress Send a key or key chord to the current controlled window. Modifier keys: cmd/command on macOS, win/windows on Execute mention_agent Mention another specialized agent to request help, delegate work, request a review, or share information. Execute monty_run Run deterministic, sandboxed Python-subset code (Monty) for fast local computation and post-processing. Execute monty_run_transform Run a named workspace transform from .cowork/transforms/<name>.monty. The input object is available as Execute monty_transform_file Read a text file, run a named workspace transform, and write the result to an output file, returning only a sm Execute move_mouse Move the pointer inside the current controlled window using screenshot-relative coordinates. Returns a fresh s Execute node_camera_snap Take a photo using a mobile companion Execute node_sms_send Send an SMS message using an Android mobile companion device. Not available on iOS. Execute open_application Open an application by name (e.g., Execute open_path Open a file or folder with the system default application Execute open_url Open a URL in the default web browser Execute orchestrate_agents Spawn multiple sub-agents in parallel and wait for all of them to complete. Execute pause_agent Pause a running descendant child agent task. This tool only works for tasks spawned by the current task (desce Execute qa_check Run a specific QA check on the current page. Execute qa_interact Perform an interaction on the current QA page (click, fill, hover, scroll, wait). Execute qa_navigate Navigate the QA browser to a URL and get a page state report. Execute qa_run Run a full automated QA pipeline on a web app using Playwright. Execute resume_agent Resume a paused descendant child agent task. This tool only works for tasks spawned by the current task (desce Execute rhino.generate_massing Generate building massing from the brief and constraints Execute run_applescript Execute exact AppleScript / osascript code on macOS. Execute run_command Execute a shell command in the workspace directory. IMPORTANT: This tool requires user approval before executi Execute schedule_task Schedule a task to run at a specific time or on a recurring basis. Execute scrape_session Run a multi-step scraping session with persistent browser state. Execute scroll Scroll at a screenshot-relative point inside the current controlled window. scrollX and scrollY are signed lin Execute show_in_folder Show a file in the system file manager (Finder on macOS, Explorer on Windows) Execute Skill Invoke a skill by ID when one clearly matches the task. If a matching skill exists, call Skill before using ot Execute spawn_agent Spawn a new agent (sub-task) to work on a specific task independently. Use this to delegate work, Execute switch_workspace Switch to a different workspace/working directory. Use this when you need to work in a different folder Execute type_text Type or set text into the currently focused control of the current controlled window. Returns a fresh screensh Execute voice_call Initiate an outbound phone call via ElevenLabs Agents + Twilio integration. Execute wait Pause briefly and then refresh the current controlled window screenshot. Useful after async UI transitions. Execute wait_for_agent Wait for a spawned agent to complete and retrieve its results. Returns the agent
WRITE 120 tools
Write acknowledge_mention Acknowledge a mention, indicating you have seen it and will work on it. Write add_behavioral_rule Add a behavioral rule that shapes how the assistant responds. Use when the user wants explicit do/don Write apple_calendar_action Use the local Apple Calendar app on macOS to list and manage events. Write apple_reminders_action Use the local Apple Reminders app on macOS to list and manage reminders. Write batch_image_process Batch process multiple images: resize, convert format, and/or add a watermark/logo. Write blender.assign_materials_by_layer Assign Blender materials using layer/object names Write blender.create_material Create or update one Blender material Write blender.create_scene Create a new Blender scene for an architecture project Write blender.import_model Import a Rhino-exported model into Blender Write blender.open_scene Open an existing Blender scene Write blender.save_scene Save the active Blender scene Write blender.set_render_engine Set render engine and quality options Write blender.setup_camera Create or update a render camera Write blender.setup_lighting Create or update scene lighting Write box_action Use the connected Box account to search, read, and manage files/folders. Write browser_save_pdf Save the current page as a PDF Write browser_upload_file Upload a workspace-readable file into a file input, preferably using a Browser V2 ref from browser_snapshot Write calendar_action Use the connected Google Calendar account to list and manage events. Write canvas_checkpoint Save a named checkpoint of the current canvas state. Write canvas_close Close a canvas session and its window Write canvas_create Create a new Live Canvas session for displaying interactive HTML/CSS/JS content. Write canvas_hide Hide the canvas window without closing the session Write canvas_push Push HTML/CSS/JS content to a canvas session. Write canvas_restore Restore a canvas session to a previously saved checkpoint. Write cloud_sandbox_write_file Write a file to a cloud sandbox. Write comfyui.collect_outputs Collect output image URLs for a completed prompt id and optionally copy them under the project root Write complete_mention Mark a mention as completed after finishing the requested work. Write copy_file Copy a file to a new location. Supports binary files (DOCX, PDF, images, etc.) and preserves exact file conten Write create_diagram Create and display a Mermaid diagram in the UI. Use this to visualize workflows, Write create_directory Create a new directory Write create_document Create a Word document (.docx) or PDF. Only use when the user EXPLICITLY requests Word/DOCX/PDF format. For al Write create_issue Create a new issue in the CoWork OS control plane. Write create_presentation Create a PowerPoint presentation Write create_spreadsheet Create an Excel spreadsheet with data, formulas, and formatting Write domain_dns_add Add a DNS record to a domain. Write dropbox_action Use the connected Dropbox account to search, read, and manage files/folders. Write edit_document Edit an existing Word document (DOCX). Supports append, move_section, insert_after_section, replace_blocks, an Write edit_file Perform surgical text replacements in files. Write edit_pdf_region Edit a selected region in an existing PDF and write the result to a new PDF path while preserving the source f Write generate_document Generate a styled PDF document from markdown content or structured sections. Write generate_epub Generate an EPUB ebook from chapter content. Write generate_landing_page Generate a polished standalone HTML landing page. Write generate_presentation Generate a PowerPoint (PPTX) presentation from structured slide data. Write generate_spreadsheet Generate an Excel (XLSX) spreadsheet from structured data with headers and rows. Write git_commit Commit changes in the workspace. Only available when working in an isolated worktree branch. Stages all change Write gmail_action Use the connected Gmail account to search, read, draft, label, archive, and send messages. Write google_drive_action Use the connected Google Drive account to search, read, and manage files/folders. Write google-workspace.calendar_event_create Create a Google Calendar event. Confirm with the user before calling. Write google-workspace.calendar_event_update Update a Google Calendar event. Confirm with the user before calling. Write google-workspace.chat_messages_create Send a message to a Google Chat space Write google-workspace.docs_append_text Append plain text to the end of a Google Document Write google-workspace.docs_create Create a new Google Document Write google-workspace.sheets_create Create a new Google Spreadsheet Write google-workspace.sheets_values_append Append rows to a spreadsheet after the last row with data Write google-workspace.sheets_values_update Write values to a spreadsheet range Write google-workspace.slides_add_text_box Add a text box to a Google Slides slide Write google-workspace.slides_create Create a new Google Slides presentation Write google-workspace.slides_create_slide Create a slide in a Google Slides presentation Write google-workspace.slides_replace_all_text Replace matching text throughout a Google Slides presentation Write google-workspace.tasks_complete Mark a Google Tasks task completed Write google-workspace.tasks_create Create a task in a Google Tasks task list Write google-workspace.tasks_lists_create Create a Google Tasks task list Write google-workspace.tasks_lists_update Update a Google Tasks task list title Write google-workspace.tasks_move Move a Google Tasks task within its list Write google-workspace.tasks_uncomplete Mark a Google Tasks task as needsAction Write google-workspace.tasks_update Update a Google Tasks task with provided fields Write integration_setup Inspect, list, or configure integrations directly from chat. Write kg_add_observation Add a timestamped observation or fact to an entity. Observations are append-only notes that track changes, dis Write kg_create_edge Create a typed relationship (edge) between two entities in the knowledge graph. Common edge types: uses, depen Write kg_create_entity Create or update an entity in the workspace knowledge graph. Entities represent people, projects, technologies Write kg_invalidate_edge Invalidate an active relationship without deleting its history. Use this when an older fact should stop being Write kg_update_entity Update an existing entity Write link_project_workspace Link a workspace to a project in the CoWork OS control plane database. Write mailbox_action Unified Inbox Agent workflow over Gmail or IMAP mailboxes: sync, score threads, summarize, draft replies, extr Write manage_heartbeat Enable or disable the heartbeat (periodic wake-up) for a digital twin / agent role. Write memory_curate Manage the small curated hot-memory layer that is injected by default. Write memory_save Save an insight, decision, observation, or error to the workspace memory database. Write notion_action Use the connected Notion account to search, read, and update pages/data sources. Write onedrive_action Use the connected OneDrive account to search, read, and manage files/folders. Write organize_folder Organize files in a folder by type, date, or custom rules Write qa_cleanup Tear down the QA browser and dev server. Write rename_file Rename or move a file Write request_user_input Ask the user a structured multiple-choice question set and block until they respond. Write revise_plan Revise the execution plan. Use this when you encounter unexpected obstacles, Write rhino.create_project Create a Rhino architecture project workspace Write rhino.create_setback_curves Create zoning/setback curves Write rhino.create_site_terrain Create terrain/topography from site parameters Write rhino.export_model Export model for Blender or delivery Write rhino.generate_floor_plan Generate floor plan rooms, circulation, and labels Write rhino.import_site_image Import a site image/reference into the Rhino project Write rhino.open_project Open an existing Rhino model or project file Write rhino.place_doors_windows Place doors/windows/openings in the active model Write rhino.place_pool Place or revise pool geometry Write rhino.save_project Save the active Rhino project Write scrape_page Scrape a web page with advanced anti-bot bypass and stealth capabilities. Write scratchpad_write Write a note to your session scratchpad. Use this to record progress, Write send_agent_message Send a follow-up message to a descendant child agent task. Use this to clarify instructions, provide missing Write set_agent_name Set or change the assistant Write set_expertise Set the assistant Write set_persona Change the assistant Write set_quirks Set personality quirks like catchphrases, sign-offs, or analogy themes. Use when the user wants the assistant Write set_response_style Adjust how the assistant responds. Use when the user asks for different response styles like Write set_user_name Store the user Write set_vibes Update the current workspace vibes/energy mode. Call this when you detect a shift in the user Write sharepoint_action Use the connected SharePoint account to search sites and manage drive items. Write skill_create Create a new custom skill. The skill will be saved to the managed skills directory Write skill_duplicate Duplicate an existing skill with a new ID and optional modifications. Write skill_proposal Manage approval-gated skill proposals. Create proposals for missing capabilities, list pending proposals, eval Write skill_update Update an existing skill. Only managed and workspace skills can be updated (not bundled). Write supermemory_remember Create a durable memory directly in Supermemory for this workspace/container. Write task_list_create Create the initial ordered session checklist for non-trivial execution work that changes artifacts/state or sp Write task_list_update Replace the full ordered session checklist state. Existing items keep their ids when supplied; new items may o Write terminate_macos_app_processes Terminate running macOS processes matching an app name or bundle-related query without shell pipelines. Write update_lore Record a notable shared moment or reference in the workspace lore. Use after significant accomplishments, Write visual_open_annotator Open an interactive visual annotation UI in Live Canvas for a workspace image. Write visual_update_annotator Update an existing visual annotation canvas session with a new image iteration (e.g., after regenerating). Write write_clipboard Write text to the system clipboard Write write_file Write content to a file in the workspace (creates or overwrites). For temporary scratch files, repro scripts, Write x_action Use the connected X/Twitter account to read, search, and post. Write youtube_ingest_video Ingest a public YouTube video
READ 131 tools
Read acp_discover Discover ACP/A2A agents that CoWork can delegate work to. Use this before selecting acp_agent_id for spawn_age Read analyze_image Analyze an image file from the workspace using a vision-capable LLM. Read blender.capture_viewport Capture the current Blender viewport Read blender.health Check Blender bridge availability Read browser_console Return recent browser console messages with secrets redacted Read browser_downloads Return recent downloads observed in the browser session Read browser_get_content Get the text content, links, and forms from the current page. Read browser_get_text Get the text content of an element Read browser_network Return recent browser network requests/responses with secrets redacted Read browser_screenshot Take a screenshot of the current page Read browser_snapshot Get a compact accessibility snapshot of the current browser tab. Prefer refs from this snapshot for browser_cl Read browser_storage Return redacted local/session storage for the current page Read browser_tabs List tabs for the active browser session Read canvas_checkpoints List all saved checkpoints for a canvas session Read canvas_list List all active canvas sessions for the current task Read canvas_snapshot Take a screenshot of the canvas content. Read capture_agent_events Capture recent events/output from a descendant child agent task. Returns a compact, summarized event list. Read channel_download_discord_attachment Download all attachments from a specific Discord message to the local inbox. Read channel_fetch_discord_messages Fetch recent messages directly from a Discord channel via the live API (not the local gateway log). Read channel_history Fetch recent message history for a specific chat from the local gateway message log. Read channel_list_chats List recently active chats for a given messaging channel (from the local gateway message log). Read cloud_sandbox_list List all active cloud sandboxes. Read cloud_sandbox_read_file Read a file from a cloud sandbox. Read cloud_sandbox_url Get the public URL for an exposed port on a cloud sandbox. Read comfyui.get_history Get ComfyUI history, optionally for one prompt id Read comfyui.get_job_status Get queue/status for a submitted prompt Read comfyui.health Check ComfyUI API availability Read comfyui.list_workflows List local workflow JSON files from COMFYUI_WORKFLOW_DIR Read count_text Count text length and basic units exactly (characters, words, lines, paragraphs, sentences). Read demo_read Demo read tool Read domain_dns_list List DNS records for a domain. Read domain_list List all domains registered on the configured account. Read domain_search Search for available domains. Returns availability and pricing for requested TLDs. Read email_imap_unread Fetch unread emails directly from the configured Email channel mailbox (IMAP/SMTP or LOOM mode). Read extract_json Extract and parse JSON objects/arrays from messy text (e.g., model output with prose and code fences). Returns Read find_macos_app_processes Find running macOS processes matching an app name or bundle-related query without shell pipelines. Read get_agent_status Check the status of spawned agents. Returns current status, progress, and any results. Read get_app_paths Get common system paths (home, downloads, documents, desktop, temp) Read get_current_location Get the user Read get_env Get the value of an environment variable Read get_file_info Get file or directory metadata (size, timestamps, permissions) Read get_orchestration_status Get the persisted status of a DAG orchestration run created for the current task. Read get_pending_mentions Get any pending mentions directed to this agent. Read get_video_generation_job Check the status of an async video generation job. Call this after generate_video returns pending=true. Return Read git_diff Show the diff of changes in the workspace. Can show unstaged changes, staged changes, or changes to a specific Read git_status Show the current git status of the workspace (changed files, staged files, branch info). Use this to understan Read glob Fast file pattern matching tool. Use glob patterns like Read google-workspace.calendar_availability_get Look up busy windows for one or more Google calendars before scheduling Read google-workspace.calendar_calendars_list List calendars visible to the authenticated Google Calendar user Read google-workspace.calendar_event_get Get a Google Calendar event by ID Read google-workspace.calendar_events_batch_get Get multiple Google Calendar events by ID from one calendar Read google-workspace.calendar_events_list List or search Google Calendar events in a calendar Read google-workspace.chat_messages_list List messages in a Google Chat space Read google-workspace.chat_spaces_list List Google Chat spaces (rooms and direct messages) the authenticated user belongs to Read google-workspace.docs_get Get a Google Document including its full content and structure Read google-workspace.drive_files_get Get metadata for a specific Drive file Read google-workspace.drive_files_list List or search files in Google Drive using Drive query syntax Read google-workspace.health Check Google Workspace connector health and authentication status Read google-workspace.sheets_get Get spreadsheet metadata including sheet names, dimensions, and properties Read google-workspace.sheets_values_get Read cell values from a spreadsheet range (e.g. Read google-workspace.slides_get Get a Google Slides presentation including page structure Read google-workspace.tasks_get Get one Google Tasks task Read google-workspace.tasks_list List tasks in a Google Tasks task list Read google-workspace.tasks_lists_list List Google Tasks task lists Read grep Powerful regex-based content search across files. Read infra_status Get overall infrastructure status: provider connections, active sandboxes, wallet state. Read kg_get_neighbors Get entities connected to a given entity via relationships. Supports multi-hop traversal up to 3 hops deep. Read kg_get_subgraph Get a subgraph containing specified entities and all edges between them. Useful for understanding how a set of Read kg_search Search for entities in the knowledge graph by name or description. Uses full-text search with ranking. Read list_agent_roles List all available specialized agent roles that can be mentioned. Read list_agents List all spawned child agents for the current task. Shows their status, model, title, and progress. Read list_directory List files and folders in a directory Read list_directory_with_sizes List files and folders in a directory with size summary (MCP-style output) Read list_goals List all goals in the CoWork OS control plane. Read list_issues List issues in the CoWork OS control plane with optional filters. Read list_macos_launch_agents List macOS LaunchAgents/LaunchDaemons matching an app query without shell pipelines. Read list_projects List all projects in the CoWork OS control plane. Read list_workspaces List all workspaces registered in CoWork OS. Read maps.health Check maps connector provider configuration Read maps.place_details Fetch normalized details for a place returned by maps.search_places Read maps.rank_nearby_options Search nearby places and rank options by walking time and relevance for urgent errands Read maps.route Estimate a walking route between two coordinates Read maps.search_places Search nearby places for a query around a latitude/longitude Read memory_curated_read Read the current curated hot-memory layer. Use this when you need to inspect what will be injected by default. Read memory_details Fetch full structured details only for selected memory IDs after memory_search_index or memory_timeline narrow Read memory_search_index Search structured memory observations and return a compact index first. Prefer this over broad memory detail r Read memory_timeline Load compact observations around a memory ID or the best match for a query before requesting full details. Read monty_list_transforms List workspace-local Monty transforms from .cowork/transforms/.monty with basic metadata (name, description, f Read node_describe Get detailed information about a connected mobile companion device, including its capabilities, permissions, a Read node_list List connected mobile companion devices (nodes). Returns information about each connected device including its Read node_location Get the current GPS location from a mobile companion device. Returns latitude, longitude, and accuracy. Read node_screen_record Record the screen of a mobile companion device. Returns the video as base64-encoded MP4. The node app must be Read parse_document Read and extract text from a local document file. Supports PDF, DOCX, XLSX, PPTX, CSV, JSON, and Markdown. Read qa_report Get the full QA run report including all checks, issues, screenshots, and interaction log. Read qa_screenshot Take a QA screenshot of the current page with diagnostics. Read read_clipboard Read the current contents of the system clipboard Read read_file Read the contents of a file in the workspace. Supports plain text files, DOCX (Word documents), PDF, and PPTX. Read read_files Read multiple files in one call using glob patterns. Useful for quickly attaching context without many read_fi Read read_pdf_visual Visually analyze a PDF document Read resolve_app_bundle_id Resolve an installed macOS app name or existing bundle identifier to the exact bundle identifier. Read rhino.capture_viewport Capture the active Rhino viewport to an image Read rhino.health Check Rhino bridge availability Read rhino.read_layers Read Rhino layer/object summary Read rhino.validate_model Validate layer, geometry, and architectural constraints Read scrape_extract Extract structured data from a web page. Automatically detects and extracts Read scrape_multiple Scrape multiple URLs in a single operation. Efficient for batch extraction Read scraping_status Check if the Scrapling library is installed and available. Read scratchpad_read Read notes from your session scratchpad. Returns all notes if no key specified, Read screen_context_resolve Resolve vague on-screen references like Read screenshot Capture the current controlled window in a native desktop app. Call this first. Read search_files Search for files by name or content Read search_memories Search the workspace memory database for past observations, decisions, and insights Read supermemory_profile Fetch the current external Supermemory profile for this workspace or container. Read supermemory_search Search Supermemory for relevant external memories and indexed chunks. Read system_info Get system information including OS, CPU, memory, and user details Read take_screenshot Take a screenshot of the screen and save it to the workspace Read task_events Query task event logs (tool calls, tool results, assistant/user messages, feedback, file ops) from the local d Read task_history Query your recent task history and messages from the local database. Read task_list_list Return the current session checklist state, including whether a verification nudge is active. Read text_metrics Compute comprehensive text metrics (characters, words, lines, paragraphs, sentences). Read tool_search Search deferred or specialized tools by intent. Use this when the exact tool name is unknown, Read wallet_balance Get the current USDC balance of the infrastructure wallet. Read wallet_info Get wallet address, network, and USDC balance. The wallet is used for infrastructure payments. Read web_fetch Fetch and read content from a SPECIFIC URL. PREFERRED for reading a known page. Returns the page content as re Read web_search Search the web for information. This is the PRIMARY tool for research tasks - finding news, trends, discussion Read x_search Search X (Twitter) posts, profiles, and threads using xAI Read x402_check Check if a URL requires x402 payment. Returns payment details if 402 status is returned. Read youtube_ask_or_ingest_video Ask a question about a public YouTube video, ingesting captions first if needed. This may access the network t Read youtube_ask_video Ask a question against locally ingested YouTube transcripts without network access. Returns an extractive answ Read youtube_list_ingested_videos List recently ingested YouTube videos in the local CoWork OS cache. Read youtube_search_ingested_segments Search locally ingested YouTube transcript segments and return timestamped matching moments.

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

Questions about CoWork OS

Can an AI agent move money through the CoWork OS MCP server? +

Yes. The CoWork OS server exposes 2 financial tools including domain_register, x402_fetch. Without a policy, an autonomous agent can call these with no spend caps, no rate limits, and no approval flow. PolicyLayer lets you block financial tools by default, require human approval, or set per-tool rate limits — enforced on every call.

Can an AI agent delete data through the CoWork OS MCP server? +

Yes. The CoWork OS server exposes 15 destructive tools including cancel_agent, cancel_video_generation_job, cloud_sandbox_delete. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through CoWork OS? +

The CoWork OS server has 120 write tools including acknowledge_mention, add_behavioral_rule, apple_calendar_action. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach CoWork OS.

How many tools does the CoWork OS MCP server expose? +

342 tools across 4 categories: Destructive, Execute, Read, Write. 131 are read-only. 211 can modify, create, or delete data.

How do I enforce a policy on CoWork OS? +

Register the CoWork OS MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every CoWork OS tool call.

Deterministic rules across all 342 CoWork OS tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

Instant setup, no code required.

342 CoWork OS tools catalogued and risk-classified — across an index of 46,500+ MCP servers.

// WHERE THIS COMES FROM

These policies come from CoWork OS's registry record.

The record behind this page: verified identity, auth posture, risk grade, every tool classified, recommended policy — re-checked continuously.

Teams ship this data inside their own products. See what a licence covers →

// GET IN TOUCH

Have a question or want to learn more? Send us a message.

Message sent.

We'll get back to you soon.