Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.praxis-ai.com/llms.txt

Use this file to discover all available pages before exploring further.

The Tools tab allows you to manage the AI tools available in your Digital Twin. This includes enabling or disabling specific tools that assist the AI in performing various tasks. Your Digital Twin uses tools to assist with specific functionality—memory, image analysis and generation, web browsing, document retrieval, LMS integration, and more. Use this page to select which tools you wish to disable (if any) for your instance. By default, all applicable tools are enabled. MCP Servers and Tools

​
Per-Assistant Tool Control

In addition to institution-level tool management, you can configure tool access per assistant. Each assistant can be set to use all tools (default) or a curated whitelist of specific tools. See Assistants — Per-Assistant Tool Configuration for details.

​
Tools Reference

This is the complete list of built-in tools available to your Digital Twin. The AI automatically selects which tool(s) to use based on the conversation context.
Some tools require additional configuration (e.g., call_canvas requires Canvas LMS integration, search_kaltura requires Kaltura credentials). Tools without required configuration are automatically disabled.

​
Web & Research Tools

Searches trusted sites with Serp APIsUse for up-to-date information on broad topics when the exact URL is unknown:
  • Current events, weather, stock tickers, sports scores
  • General web research not covered by specialized tools
  • Information you’re unfamiliar with
Examples:
  • “Search for the latest news on climate change.”
  • “What’s the current weather in Tokyo?”
  • “Find the most popular tourist attractions in Paris.”
Supports real-time voice conversations.
Read content at URLs on the internetAccess, retrieve, summarize, or save content from a known web page URL. Provides download links in Markdown format for public files.Examples:
  • “Read the content from the Wikipedia page on the Roman Empire.”
  • “Summarize the main points from https://example.com/article”
  • “Save the content from this GitHub README to my files.”
Supports real-time voice conversations.
Searches Google ScholarSearch for scholarly articles and publications from peer-reviewed sources including NIH, NIM, PubMed, DOI, and more.Examples:
  • “Find scholarly articles on artificial intelligence in healthcare.”
  • “Search for recent studies on the effects of social media on mental health.”
  • “What does the research say about intermittent fasting?”
Searches Google PatentsSearch for information about patents and inventions worldwide.Examples:
  • “Who owns the patent for the smartphone?”
  • “Find patents related to virtual reality technology filed in the last five years.”
  • “Search for patent ID US10123456.”
Searches ClinicalTrials.govSearch clinical research studies worldwide from the US National Library of Medicine, including purpose, recruitment status, eligibility criteria, and locations.Examples:
  • “Search for clinical trials related to diabetes treatment.”
  • “Find clinical studies on mindfulness meditation for stress reduction.”
  • “What’s the status of trial NCT06310070?”
Searches WolframAlpha for computations and visualizationsFactual lookups, quantitative analysis, symbolic/numeric computation (integrals, sequences, financial math), and visualizations across science, math, geography, economics, weather, population, music, and space. Also supports bio/chem/finance tasks (e.g., DNA→amino acids translation, Taylor series, compound interest calculations).Response Styles:
  • short — Terse answer only (minimal prose)
  • image — Return response as an image (best for complex graphs)
  • full — Detailed steps/derivations with richer context (default)
  • spoken — Phrasing suitable for voice narration
Examples:
  • “What’s the nutritional content of an apple?”
  • “Calculate the integral of sin(x) from 0 to π.”
  • “Weather in NYC on July 4, 1776.”
  • “California population from 1900 to 2020.”
  • “Future value of $1,000 at 5% for 10 years.”
  • “Translate this DNA sequence to amino acids.”
Supports real-time voice conversations.

​
Media & Content Tools

Searches Google YouTubeSearch for video content on specific topics from YouTube and other sources.Examples:
  • “Find an educational video about the solar system.”
  • “Show me a tutorial on how to bake sourdough bread.”
  • “Search for analysis videos on current economic events.”
Supports real-time voice conversations.
Transcribe a video from YouTubeTranscribe video content when the full YouTube URL is known. Does not work for non-YouTube videos or audio files.Parameters:
  • URL: The YouTube video URL (e.g., https://www.youtube.com/watch?v=... or https://youtu.be/...)
  • Save Result: Set to true to save/download the transcript to your files
Examples:
Transcribe an audio file from a URLTranscribe audio files from a URL. Does not work for YouTube videos. The AI checks if transcripts are available in your files before using this tool.Supported Formats: mp3, mp4, mpeg, mpga, m4a, wav, webmExamples:
Search Kaltura for instructional videosSearch or list instructional videos on the Kaltura video streaming platform. Supports filtering by author and extracting transcripts.
Requires Kaltura integration to be configured.
Parameters:
  • Topic: The topic to search (leave empty for all videos)
  • Author: Filter by author’s user ID, name, or email (comma-separated)
  • Extract Transcripts: Set to true to include video transcripts
  • Save Results: Set to true to download search results
Examples:
  • “Find an Algebra video on Kaltura.”
  • “What are my videos on Kaltura?”
  • “Download the transcripts of a video about genome sequencing from Kaltura.”
  • “Find videos from hugo@praxis-ai.com on Kaltura.”
Supports real-time voice conversations.
Search for images using Unsplash.orgSearch for free images in the Unsplash database. Returns images with courtesy notices for attribution.Parameters:
  • Description: A short description of the image to search
  • Orientation: landscape, squarish, or portrait (default)
  • Count: Number of images to return (default: 1)
Examples:
  • “Find a sunset over the mountains.”
  • “Search for 3 landscape photos of the ocean.”
  • “Find a portrait-oriented image of a forest.”
Supports real-time voice conversations.
Search for sounds using Freesound.orgSearch for free sounds in the Freesound database. For sound effects only, not music.Parameters:
  • Search Terms: Space-separated terms, or phrase in double quotes. Use + for mandatory and - for prohibited terms (e.g., "thunder -rain")
  • Count: Number of sounds to return (default: 1)
Examples:
  • “Find a sound of whales talking to each other.”
  • “Search for thunder and rain sounds.”
  • “Find 5 different bird call sounds.”
Supports real-time voice conversations.
Get address from GPS coordinatesObtain the address of a place based on latitude and longitude coordinates.Examples:
  • “What’s the address at coordinates 40.7128, -74.0060?”
  • “What city is at latitude -21.2863697 and longitude 55.4108081?”
Supports real-time voice conversations.

​
Image Tools

Generate or edit images using AICreate new images or edit existing ones using the configured Image Generation Model (e.g., DALL-E, Amazon Titan). Prompts are automatically translated to English. Supports generating multiple variations from the same prompt.Parameters:
  • Prompt: Detailed description of the image to create
  • Source URLs: Optional space-delimited list of image URLs to edit/extend (JPEG, PNG, WebP only)
  • Count: Number of image variations to generate (default: 1)
Rules:
  • No politicians or public figures
  • No copyrighted characters (original descriptions only)
  • Supports editing JPEG, PNG, and WebP files
Examples:
  • “Generate an image of a serene lake at sunset with mountains.”
  • “Create a 3D picture of DNA proteins with bright colors.”
  • “Edit this image to add a rainbow in the sky.”
  • “Generate 3 variations of a futuristic cityscape.”
Supports real-time voice conversations.
Analyze and describe content in imagesIdentify, describe, extract text, count elements, determine emotions, historical context, cultural significance, and accessibility options in images.Examples:
  • “Analyze this image of the Eiffel Tower for historical context.”
  • “Extract the text contained in this screenshot.”
  • “Describe the species of birds in this photo.”
  • “Count the number of people in this image.”
Supports real-time voice conversations.

​
Document & File Tools

Retrieve content from user files via semantic searchSearch for content relevant to the conversation contained in user files using Euclidean kNN (k-Nearest Neighbor) search. Results are ordered by relevance score.Examples:
  • “Search my documents for information about project budgets.”
  • “Find content in my files related to marketing strategies.”
  • “What do my uploaded documents say about compliance requirements?”
Supports real-time voice conversations.
Retrieve user files with filteringAccess the user’s document library with metadata including summaries, types, dates, dimensions, and authors. Filter by file type or keyword search.Parameters:
  • Type Filter: Filter by image, audio, video, or document (for all other files)
  • Keyword: Search within file names and document summaries (exact match)
  • Include Content: Set to true to include file content in response
Examples:
  • “List all my uploaded documents.”
  • “Show me my image files.”
  • “Find files with ‘report’ in the name.”
  • “What audio files do I have?”
  • “Show me the contents of my research documents.”
Supports real-time voice conversations.
Save text content to user’s document collectionSave or append text content to files. Supports text-based formats only (MD, TXT, JSON, CSV, HTML, JS, CSS, XML, YAML, TEX, TSV). Binary formats (PDF, DOCX) are not supported.Parameters:
  • Content: The text content to save (required)
  • Filename: File name without extension. Auto-generated if not specified using format {subject}_{YYYY_MM_DD_HH_MM}
  • Extension: File extension without leading dot (default: txt)
  • Overwrite: Set to true to replace existing file, false to append (default)
Examples:
  • “Save this conversation summary to a file called meeting_notes.”
  • “Create a new markdown file with this content.”
  • “Append these notes to my existing research.txt file.”
  • “Save this code snippet as a JavaScript file.”
Supports real-time voice conversations.

​
Memory Tools

Save parameters to user or shared memoryStore or clear parameters in memory. Each parameter includes a namespace (for categorization), name, value, and description. Use namespaces to organize parameters and avoid key name conflicts. Shared memory is accessible by all users of the Digital Twin.Parameters:
  • Parameters Array: JSON array with key_name, key_value, key_description, and key_namespace properties
  • Shared Memory: Set to true to save in shared memory (accessible by all users)
  • Clear Memory: Set to true to remove all parameters from memory
Examples:
  • “Remember that my birthday is December 30th, 1972.”
  • “Remember I prefer short and concise responses.”
  • “Save this copyright notice in shared memory: © 2025, Praxis-AI, Inc”
  • “Clear all my stored preferences from memory.”
Supports real-time voice conversations.
Retrieve parameters from memoryRetrieve a list of parameters from user or shared memory. Returns CSV-formatted data including namespaces, names, values, descriptions, and creation dates. Search is case-insensitive and token-based.Parameters:
  • Search Term: Search by key name, description, or namespace (leave empty for all parameters)
  • Shared Memory: Set to true to search in shared memory instead of user memory
Examples:
  • “What’s my morning routine according to memory?”
  • “What preferences have I stored?”
  • “Show me all parameters in shared memory.”
  • “What do you remember about my personal data?”
Supports real-time voice conversations.

​
Visualization Tools

Create 2D charts for data visualizationGenerate charts for visualizing single or multiple data series with categories on the x-axis. Each data series should include values corresponding to the specified categories.Parameters:
  • Title: Descriptive title for the chart
  • Categories: JSON array of x-axis category names (e.g., ['January', 'February', 'March'])
  • Data Series: JSON array of data series objects with name and data properties (e.g., [{name: 'Sales', data: [10, 8, 11]}])
Examples:
  • “Create a chart showing sales data from January to April.”
  • “Generate a chart comparing monthly rainfall to temperature.”
  • “Visualize this quarterly revenue data as a bar chart.”

​
Google Workspace Tools

Google Workspace tools require OAuth configuration. Each tool is automatically enabled or disabled based on whether the corresponding Google service is configured in your instance settings. See Google Workspace Integration for setup details.
Access Gmail to read, search, send, and manage emailsInteract with the user’s Gmail account — search inbox, read messages, send emails, manage labels, work with threads and drafts.Key Capabilities:
  • Search with Gmail syntax: from:, to:, subject:, is:unread, has:attachment, newer_than:7d
  • Send emails with HTML formatting, CC/BCC, and attachments
  • Manage labels, trash messages, work with drafts
Examples:
  • “Show me my unread emails from this week.”
  • “Search my inbox for messages from boss@company.com about invoices.”
  • “Send an email to john@example.com about the meeting tomorrow.”
  • “Find emails with PDF attachments from the last 30 days.”
  • “Move that email to trash.”
The AI always confirms with the user before sending emails.
Search, browse, and download files from Google DriveSearch and list files in the user’s Google Drive, get file metadata, and download files to the user’s IP Vault for RAG indexing.Key Capabilities:
  • Search by name, type, date, or full-text content
  • Download files to IP Vault (triggers RAG indexing)
  • Export Google Workspace files (Docs, Sheets, Slides) to PDF/DOCX/XLSX/PPTX
  • View file permissions and comments
Examples:
  • “Search my Google Drive for the Q4 report.”
  • “Download the project proposal to my files.”
  • “Find all presentations modified in the last week.”
  • “Show me my Drive storage usage.”
Supports real-time voice conversations. To read file content directly, use the dedicated Docs, Sheets, or Slides tools instead of downloading.
List, create, update, and delete calendar eventsFull read/write access to the user’s Google Calendar with timezone-aware event management, recurring events, and availability checks.Key Capabilities:
  • List events for any date range
  • Create events with attendees, location, and recurrence
  • Check free/busy availability
  • Update or delete existing events
Examples:
  • “What’s on my calendar for tomorrow?”
  • “Schedule a team meeting next Tuesday at 2 PM for 1 hour.”
  • “Create a recurring standup every weekday at 9 AM.”
  • “Am I free this Friday afternoon?”
  • “Cancel my 3 PM meeting.”
The AI confirms with the user before creating, modifying, or deleting events.
Read and analyze spreadsheet data directlyRead Google Sheets content directly from Google’s API — no download needed. Supports A1 notation for cell ranges, formulas, and structured table data.Key Capabilities:
  • Read specific cell ranges using A1 notation (e.g., Sheet1!A1:D10)
  • View formatted values, raw values, or formulas
  • Get spreadsheet metadata and sheet names
Examples:
  • “Read the data from my Budget spreadsheet.”
  • “Show me cells A1 through D20 from Sheet1.”
  • “What formulas are used in the Revenue column?”
  • “Summarize the data in my Sales Tracker spreadsheet.”
Requires Google Drive to be enabled (dependency).
Read and analyze document content directlyRead Google Docs content directly from Google’s API — no download needed. Extracts plain text with table support for summarization and analysis.Examples:
  • “Summarize my Google Doc about marketing strategy.”
  • “Read the content of my project proposal document.”
  • “What does my meeting notes document say about action items?”
Requires Google Drive to be enabled (dependency).
Read and analyze presentation content directlyRead Google Slides content directly from Google’s API — no download needed. Extracts text from all slides with slide-by-slide structure.Examples:
  • “What’s in my Google Slides presentation about onboarding?”
  • “Summarize the key points from my quarterly review slides.”
  • “Read slide 3 of my training presentation.”
Requires Google Drive to be enabled (dependency).
Access recordings and transcripts from past Google Meet meetingsRetrieve conference history, recordings (with Drive file links), and full transcripts with speaker attribution from past Google Meet sessions.Examples:
  • “List my recent Google Meet recordings.”
  • “Get the transcript from yesterday’s team meeting.”
  • “Find the recording from last week’s project review.”
For scheduling meetings with video links, use call_google_calendar instead. This tool is for past meeting data only. Requires Google Calendar to be enabled (dependency).
View courses, assignments, grades, and announcements from Google ClassroomRead-only access to the user’s Google Classroom — list courses, view assignments and submissions, read announcements.Examples:
  • “List my Google Classroom courses.”
  • “What assignments are due this week in Biology 101?”
  • “Show me the latest announcements from my course.”
  • “What’s my grade on the midterm submission?”
Find places, get location details, and convert between addresses and coordinatesSearch for places, get business details and reviews, perform geocoding and reverse geocoding using Google Maps.Key Actions:
  • text_search — Find places by query (e.g., “coffee shops in Seattle”)
  • nearby_search — Find places near GPS coordinates by type
  • place_details — Get hours, phone, reviews for a specific place
  • geocode / reverse_geocode — Convert between addresses and coordinates
  • get_user_location — Get the user’s current location (browser permission required)
Examples:
  • “Find Italian restaurants near downtown Seattle.”
  • “What’s the address at coordinates 40.7128, -74.0060?”
  • “Show me coffee shops within 2 km of my location.”
  • “Get reviews for this restaurant.”
Supports real-time voice conversations. Uses server-side API key — no user OAuth required.

​
LMS Integration Tools

Interact with Canvas LMSAccess course material (syllabus, pages, office hours), enrolled users, assignments, submissions, grades, announcements, and discussion topics from Canvas LMS via the Canvas API.
Requires Canvas LMS integration to be configured.
Parameters:
  • Method: HTTP request method (GET, POST, PUT, DELETE)
  • Endpoint: Canvas API endpoint URL (without domain)
  • Body: Request body for POST/PUT requests (form-urlencoded or JSON)
  • Faculty Context: Set to true for faculty/admin actions (grading, curating content)
Supported Operations:
  • List and retrieve course content, syllabus, and pages
  • View and submit assignments
  • Read and post discussion entries
  • Access grades and submission feedback
  • Smart search via /api/v1/courses/:course_id/smartsearch
Examples:
  • “What are my assignments due this week?”
  • “Show me the course syllabus.”
  • “List discussion topics in my course.”
  • “What’s my grade on the midterm exam?”
  • “Post a comment on my assignment submission.”
Supports real-time voice conversations.
Search Canvas for course materialFallback search tool when call_canvas is not available. Searches course material including syllabus, pages, content, assignments, grades, office hours, and announcements.
Requires Canvas LMS integration to be configured.
Examples:
  • “When is my homework due this week for Intro to Marketing?”
  • “Search for office hours in my course.”
  • “Find content about Chapter 5 in my course.”

​
Bioinformatics Tools

BLAST sequence similarity searchFind regions of similarity between biological sequences using NCBI BLAST (nucleotide/protein comparison with statistical significance). Note: Queries may take several minutes.Parameters:
  • Program: megablast, blastn, blastp, rpsblast, blastx, tblastn, or tblastx (required)
  • Database: Target database for comparison (required)
  • Sequence: Nucleotide or protein sequence to search (required)
  • Save Results: Set to true to download results
Common Databases:
  • All programs: core_nt, refseq_select_rna, refseq_rna, nt, Whole_Genome_Shotgun_contigs, est, sra, patnt, pdbnt
  • Protein (blastp, blastx): nr, refseq_select_prot, refseq_protein, swissprot, pataa, pdb
  • Nucleotide (blastn): rRNA_typestrains/16S_ribosomal_RNA, nt_euk, nt_prok, nt_viruses
Examples:
  • “Search blastp for this protein sequence: MKVKVLSLLVP…”
  • “Run a megablast search against the nt database.”
  • “Compare this DNA sequence to RefSeq genomes.”

​
Communication Tools

Send an emailSend emails with specified recipients, subject line, and HTML-formatted message body. Supports multiple recipients and responsive HTML formatting.
The AI will request confirmation before sending unless instructed otherwise in your Digital Twin’s system prompt.
Parameters:
  • Recipients: Email address(es), comma-separated for multiple recipients (required)
  • Subject: Email subject line (required)
  • Body: HTML-formatted message with inline CSS and responsive tags like <p>, <br>, <strong> (required)
Examples:
Supports real-time voice conversations.

​
Assistant Tools

List available assistantsRetrieve a filtered or complete list of conversational assistants accessible to the current user. Returns assistant IDs and names.Examples:
  • “What assistants are available?”
  • “Find assistants with ‘math’ in the name.”
  • “List all my assistants.”
Supports real-time voice conversations.
Activate an assistantInitialize and activate a specific conversational assistant, loading its specialized instructions into the current session.Examples:
  • “Start the Math Tutor assistant.”
  • “Switch to the Writing Coach.”
  • “Activate the Code Review assistant.”
Available in real-time voice mode only.

​
API Integration Tools

Execute REST API callsExecute authenticated REST API calls to specific service endpoints. Use only when: (1) explicitly instructed by name, (2) the endpoint requires authentication other tools can’t provide, or (3) the service is internal/proprietary.
This tool is for internal/proprietary services requiring authentication. Use read_url or get_browser for public web content.
Parameters:
  • Bearer Token: Authorization token (leave empty if no auth required)
  • URL: Complete REST endpoint URL with all variables substituted
  • Method: HTTP method (GET, POST, PUT, DELETE)
  • Body: Request body in JSON format for POST/PUT/DELETE
  • Query Params: URL-encoded query string for GET/DELETE (RFC 3986)
  • Content Type: Header value (default: application/json)
Common Responses: 401 (auth required), 400 (invalid params), 404 (endpoint not found), 500 (service unavailable)Examples:
  • “Call the attendance API endpoint with my bearer token.”
  • “POST this data to the internal reporting service.”
  • “Fetch data from our custom REST endpoint.”
Supports real-time voice conversations.

​
Tool Availability

Tools are automatically enabled or disabled based on your Digital Twin’s configuration:
ToolRequires
call_canvas, search_canvasCanvas LMS integration
search_kalturaKaltura credentials
search_blastInstitutional account
call_google_gmailGoogle OAuth + Gmail API enabled
call_google_driveGoogle OAuth + Drive API enabled
call_google_calendarGoogle OAuth + Calendar API enabled
call_google_sheetsGoogle OAuth + Drive API enabled
call_google_docsGoogle OAuth + Drive API enabled
call_google_slidesGoogle OAuth + Drive API enabled
call_google_meetGoogle OAuth + Calendar API enabled
call_google_classroomGoogle OAuth + Classroom API enabled
call_google_mapsGoogle Maps API key (no user OAuth)
To disable specific tools for your Digital Twin, uncheck them in the Tools section of your instance settings. Disabled tools will not be available to the AI during conversations.
Tools can also be controlled at the assistant level. When an assistant is set to “Selected Tools Only” mode, only the whitelisted tools are available during conversations with that assistant — even if the tool is enabled at the institution level. See Assistants for details.

​
Real-Time Voice Support

Tools marked with
Supports real-time voice conversations.
can be used during speech-to-speech sessions. The AI can invoke these tools while maintaining a natural voice conversation. Tools marked with
Available in real-time voice mode only.
are exclusively available during real-time voice sessions (e.g., switching assistants mid-conversation).