Tools

Tool catalog

Every native tool Ultron exposes to Claude is listed below, grouped by category. The native catalog lives in src/lib/chat-tools.ts. Provider modules (Stripe, HITL) live under src/lib/tools/providers/. External MCP servers add more tools at runtime per connected user and are documented separately under External tool servers.

Updated today

Overview

At a glance

Native total: 84 tools
Provider modules: 2 (Stripe with 7 tools, HITL with 1 tool)
Grand total: 92 tools available before external MCP
Source: src/lib/chat-tools.ts
Modules: src/lib/tools/providers/
Schema shape: Tool-call function schema
Exposure: Merged with skill allow-list and filtered by connection gate

Note

A tool listed here is not automatically available to every skill. The engine intersects the manifest tools array with the tools the user has the underlying provider connected for. Skills declare what they need; the connection gate decides what actually reaches the model.

Web and research

The two general-purpose research entry points.

Name	What it does
`web_search`	Searches the web for real-time information about companies, products, pricing, news.
`web_search_multiple`	Runs up to five parallel web searches for comparison and batch lookups.

Scraping

Headless browser and Reddit-specific fetchers.

Name	What it does
`scrape_website`	Browses any URL with Chromium and returns extracted content or a screenshot.
`browser_action`	Drives a Playwright session through navigate, click, type, screenshot, extract, wait, scroll, select.
`scrape_reddit`	Pulls Reddit posts and top comments without auth via the JSON endpoint.

People and company data

Apollo-backed enrichment and search.

Name	What it does
`enrich_lead`	Enriches a lead with email, title, social profiles, and firmographics.
`search_people`	Searches contacts by domain, title, location, and keywords.
`search_companies`	Searches companies by industry, location, and employee range.

Search, browse, and post to Reddit.

Name	What it does
`search_reddit`	Searches Reddit posts and discussions across the platform.
`get_subreddit_posts`	Lists hot, top, or new posts from one subreddit.
`get_reddit_comments`	Returns the top comments on a specific Reddit post.
`reddit_submit_post`	Submits a post to a subreddit on behalf of the connected user.
`reddit_comment_on_post`	Posts a comment on a thread on behalf of the connected user.

CRM

The native CRM. Leads, deals, activities, sequences, segments, workflows, agreements.

Name	What it does
`crm_save_lead`	Saves a lead with score, status, and a research summary.
`crm_log_activity`	Logs a note, call, email, meeting, or task against a lead.
`crm_manage_tasks`	Creates, updates, or lists tasks on a deal.
`crm_email_sequences`	Creates and manages multi-step email sequences with enrollment controls.
`crm_segments`	Manages dynamic lead segments using filters.
`crm_workflows`	Manages trigger-based CRM automations.
`crm_deduplicate`	Scans for and merges duplicate leads by email or near-match.
`crm_pipeline_velocity`	Returns conversion rates, average time per stage, and win rate.
`crm_manage_deals`	Creates, lists, and updates deals across the pipeline.
`crm_manage_agreements`	Creates, sends, and tracks agreements with e-signature.
`outreach_pipeline`	Manages the intelligence-driven outbound pipeline for customer campaigns.

Email

Gmail for personal mail, Brevo for transactional and campaign delivery.

Name	What it does
`gmail_send_email`	Sends an email from the connected Gmail account with optional attachments.
`gmail_read_emails`	Reads recent emails from the connected Gmail inbox with Gmail search syntax.
`brevo_send_email`	Sends a transactional or campaign email via Brevo with attachments and tags.
`brevo_manage_domain`	Adds, lists, verifies, or removes a sending domain in Brevo.
`brevo_dns_setup`	Auto-configures Brevo DNS records on a domain managed by Cloudflare.

Calendar

Google Calendar reads and writes, plus Calendly meeting management.

Name	What it does
`calendly_list_event_types`	Lists Calendly event types with duration, URL, and status.
`calendly_list_events`	Lists upcoming or past Calendly events with invitees and status.
`calendly_get_event_details`	Returns the full record for a specific Calendly event.
`calendly_create_link`	Creates a single-use scheduling link for an event type.
`calendly_cancel_event`	Cancels a scheduled Calendly event.
`gcal_list_events`	Lists upcoming Google Calendar events for availability checks.
`gcal_create_event`	Creates a new event with attendees, location, and description.
`gcal_delete_event`	Deletes or cancels an event with optional notification controls.

Content

Social posts, automations, analytics, and the carousel generator.

Name	What it does
`content_create_post`	Drafts or schedules a social post for one or more platforms.
`content_list_posts`	Lists and filters posts by status, platform, date, or text match.
`content_update_post`	Updates an existing post's content, schedule, status, or platforms.
`content_delete_post`	Deletes a social post.
`content_get_analytics`	Returns impressions, engagement, and top posts.
`content_get_calendar`	Returns scheduled and published posts organized by date.
`content_list_accounts`	Lists connected social accounts with platform, status, and last sync.
`content_create_automation`	Creates a content automation rule with trigger and actions.
`content_list_automations`	Lists existing content automation rules.
`generate_carousel`	Generates a multi-slide carousel and returns the artifact plus preview URL.
`list_my_templates`	Lists personal and community carousel templates by tags and recipe.

Outreach automation

UI cards and pipeline operations for outbound campaigns.

Name	What it does
`show_sequence_builder`	Renders the interactive sequence builder card in the chat UI.

Note

outreach_pipeline belongs to the CRM cluster but participates in outreach flows too. See the CRM table above for its description.

AI assisted writing

Lightweight background model calls for cheap repurposing, replies, and briefs.

Name	What it does
`gemma_repurpose_content`	Adapts a piece of content into tweets, LinkedIn posts, newsletters, and Instagram captions.
`gemma_draft_reply`	Drafts an email reply that reads sentiment, objections, and buying signals.
`gemma_meeting_brief`	Builds a pre-call brief with talking points, questions, and landmines.

Integrations

Slack, Notion, the Google productivity suite, Microsoft, video conferencing, HubSpot, and n8n.

Name	What it does
`slack_send_message`	Sends a message to a Slack channel or user DM with markdown support.
`slack_list_channels`	Lists Slack channels the bot can see.
`slack_search_messages`	Searches Slack history for past conversations, decisions, and mentions.
`notion_action`	Searches, reads, creates, and updates Notion pages and databases.
`outlook_action`	Drives the connected Outlook account through send, search, and folder operations.
`google_drive_action`	Lists, uploads, downloads, and manages files in Google Drive.
`google_sheets_action`	Reads, batch-updates, and appends rows in Google Sheets.
`google_docs_action`	Creates, searches, and updates Google Docs.
`google_analytics_action`	Reads Google Analytics accounts, properties, and audiences.
`google_meet_action`	Creates Meet sessions and pulls recordings and transcripts.
`zoom_action`	Creates Zoom meetings, lists schedules, and pulls recordings.
`microsoft_teams_action`	Lists Teams, channels, and posts message replies.
`hubspot_search_contacts`	Searches HubSpot contacts by name, email, or company.
`hubspot_list_deals`	Lists HubSpot deals with stages, amounts, and close dates.
`hubspot_create_contact`	Creates a HubSpot contact with company and job title.
`n8n_list_workflows`	Lists all n8n workflows with name, id, and active status.
`n8n_trigger_workflow`	Triggers an n8n workflow by posting payload to its webhook URL.
`n8n_manage_workflow`	Activates or deactivates an n8n workflow.
`n8n_get_executions`	Returns recent execution history for an n8n workflow.

Brain and memory

Persisted memory and the notes surface.

Name	What it does
`search_memories`	Recalls stored memories using hybrid retrieval, then a lightweight relevance pass to pick the handful that matter for this turn.
`brain_manage_notes`	Creates, reads, updates, deletes, and AI-edits personal notes.

Sandbox execution

Tools that delegate work into an isolated per-conversation sandbox.

Name	What it does
`code_task`	Builds, runs, tests, and debugs software in the persistent cloud workspace.
`research_task`	Runs web research, document analysis, and fact-checking with citations.
`data_task`	Performs CSV, JSON, and statistical analysis inside the sandbox.
`multi_agent_task`	Decomposes a complex task into parallel sub-agents in the sandbox.
`deploy_task`	Ships software to a live deployment target: a container host, edge workers, or a hosted database.
`complex_execution`	Generic fallback for sandbox work that doesn't fit a more specific task tool.

Agents and skills

The tools that select and dispatch other tools.

Name	What it does
`run_skill`	Activates a registered skill as a sub-agent with its own tools and memory.
`discover_capability`	Searches the capability registry to pick the right skill or integration action.
`agent`	Dispatches a scoped slice of work to a typed sub-agent (explore / plan / coder / general) that runs in the container with its own tools.

System and workspace

Profile, pipeline summary, task list, calculations, documents.

Name	What it does
`lookup_leads`	Queries CRM leads by status, company, industry, or minimum score.
`get_tasks`	Reads home tab to-dos with status and due-date filters.
`complete_task`	Marks a home tab task as complete.
`get_company_profile`	Returns the user's company profile: name, description, ICP, voice, competitors.
`get_pipeline_stats`	Returns aggregated pipeline counts, scores, top industries, and activity.
`calculate`	Evaluates a math expression for burn rate, projections, and metrics.
`generate_document`	Generates a downloadable DOCX, XLSX, or PDF from a structured spec.

UX and interaction

The tool that talks to the user directly when the loop is uncertain.

Name	What it does
`ask_clarification`	Asks the user a multiple-choice question when the model is genuinely uncertain which tool to use.

Stripe provider module

Seven tools exported from src/lib/tools/providers/stripe.ts. Connected on the user's Stripe account.

Name	What it does
`stripe_list_customers`	Lists recent customers with optional email filter.
`stripe_list_payments`	Lists recent payments and charges with optional customer filter.
`stripe_list_subscriptions`	Lists subscriptions filtered by status for MRR and churn checks.
`stripe_create_payment_link`	Creates a payment link for a deal with an amount and description.
`stripe_create_invoice`	Creates and sends a Stripe invoice with line items and due terms.
`stripe_create_subscription`	Creates a recurring subscription with interval, amount, and optional trial.
`stripe_revenue_metrics`	Returns MRR, ARR, 30-day collected, outstanding, and subscription counts.

Human-in-the-loop module

One tool exported from src/lib/tools/providers/hitl.ts. The full mechanism is documented under Resume and approvals.

Name	What it does
`wait_for_approval`	Pauses the workflow and requests explicit user approval before an irreversible action.

Tip

Detailed contract for wait_for_approval (schema fields, pause_points table, the resume endpoint) lives at Resume and approvals.

Glossary

Native tool: A tool declared in src/lib/chat-tools.ts and exposed to Claude on every turn the connection gate allows.
Provider module: A self-contained tool bundle in src/lib/tools/providers/ that ships its own schemas, executors, and status labels.
External MCP tool: A tool discovered from a JSON-RPC server registered for the user. Not enumerated on this page; covered under External tool servers.
Connection gate: The runtime filter that removes any tool whose required provider is not connected for the current user.
Status label: A short verb phrase rendered in the activity sidebar while a tool is running. Defined per tool by the provider module or the native catalog.