Research at the Speed of Thought

Stop clicking through menus. Whether you're decoding headlines, researching dates, or building connections across events, just tell your AI assistant what to research — calculations, database lookups, date correlations, pattern matching, and report generation happen in seconds.

"What's the gematria connection between today's date and the top headlines?"

"Calculate the duration from 9/11 to today and find all matching words in my database"

"Generate a full decode card for Trump with today's date numerology"

Why Use a Gematria AI Agent?

Gematria research is powerful but time-consuming. Calculating values across multiple ciphers, cross-referencing dates, searching databases for matches, and finding patterns can take hours of manual work.

The Gematria AI Agent collapses that into seconds — and it is not a chatbot with pre-written answers. It executes real calculations, queries real databases, and generates real analysis using the same engine that powers the web tools. Ask a question, get a full decode with pattern matches and connections you might have missed, pulled from your own research databases.

How It Works

Option A: Claude.ai (Easiest)

1

Add Connector Claude.ai → Settings → Connectors → Add custom connector

2

Enter URL Paste https://gematrialens.com/mcp and click Add

3

Authorize & Go Log in to your GematriaLens account, authorize, and start researching

No API key needed. Works with any Claude.ai plan.

Option B: Claude Desktop & Other Clients (API Key)

1

Create an API Key One-click setup in your account settings

2

Connect Your AI Add the key to Claude Desktop, ChatGPT, or any MCP client

3

Start Researching Ask questions in plain language — the AI uses your tools and databases

Claude Desktop requires Node.js (v18+) for the npx bridge command.

Research Tools at Your AI's Fingertips

Every tool available on the website is accessible through the AI Agent — plus some that work even better with AI.

Gematria Calculation

Calculate values in 80+ ciphers
Per-character breakdowns
Multi-word side-by-side comparison
Reverse number lookup

Date & Numerology

Date numerology (17 variants)
Duration between dates + numerology
Date text gematria
Decode any date's values

My Predictive Calendars

Analyze word on predictive calendar
Search dates by gematria value
Create & manage calendars
Save & browse calendar events
Eclipse correlations (2001-2100)

Daily Decode

Full daily composite analysis
Headline gematria matching
Google Trends matching
Name-to-date correlations

Full Analysis

Instant multi-word analysis
Quick pattern insights
News article decoding Premium

My Databases

Create & manage word databases
Create & manage number databases
Search your collections
Add words to your databases
Pattern matching across all DBs

My Reports, Events & Cards

Manage saved events
Browse report history
Create & share decode cards
Account & usage stats

Gematria gematria_calculate gematria_calculate_detailed gematria_compare list_ciphers number_lookup number_properties number_search_databases gematria_highlights

Date & Calendar today_numerology date_numerology decode_date calculate_date_duration date_text_gematria

My Predictive Calendars predictive_calendar_analyze predictive_calendar_search list_calendars create_calendar list_calendar_events add_calendar_event delete_calendar_event delete_calendar list_upcoming_calendar_events

Daily Decode daily_decode headlines_decode google_trends_decode check_name_against_date value_word_matches

Analysis analyze_text analyze_article

My Word Databases list_word_databases create_word_database search_word_database database_add_word

My Number Databases list_number_databases create_number_database database_add_number

My Reports list_my_reports search_my_reports get_report_config get_report_results get_report_gematria

My Decode Cards list_my_cards get_card_data create_card delete_card list_card_themes

My Events list_my_events create_my_event update_my_event delete_my_event delete_my_events_bulk

Account my_account platform_stats

Research Skills Library

A skill turns a whole research workflow into a single slash command. The examples below chain multiple GematriaLens tools automatically — copy one into your AI client, adapt it, or use it as a starting point for your own.

/decode Complete multi-tool investigation on any topic, name, or event

Never compute gematria yourself. All values must come from GematriaLens tool results only.
Call list_ciphers first and build a canonical cipher map from tool data only: {code -> name}. This map is the single source of truth for labels.
Call list_word_databases and list_number_databases to get my database IDs. Use all of them.
Expand my input into phrase variants before analyzing. For a person include the common forms (e.g. "Trump", "Donald Trump", "Donald John Trump", "President Donald Trump"); for a headline include the full sentence and the key phrase on its own. List the variants you chose and let me correct them. A value that repeats across several variants is far stronger evidence than one that appears in a single phrasing — say which is which.
Tell me which ciphers are being used. analyze_text defaults to the 9 standard ciphers, but the platform has 84 — if I want wider coverage, pass explicit cipher_codes. If the response contains _premium_ciphers_skipped, tell me which ciphers were skipped and why.
Call analyze_text with the word(s) I provide and all my database IDs. Use event_date for the date of the event or announcement being researched (default today). Use custom_dates for birth dates or significant dates of related people/things in the format ["Label, YYYY-MM-DD"] — these are the dates anniversary spans are calculated FROM, not the event date itself. Leave include_my_events: false unless I explicitly ask for my saved events — they are unrelated to most stories and only add coincidental matches.
Dates for custom_dates: You may use dates I explicitly provide, or automatically supply well-known public dates (e.g. a public figure's birthday) when I ask for a span or when directly relevant to the subject. Clearly list the date and label you used (e.g. "Donald Trump Born, 1946-06-14") and state it for confirmation so I can correct it if needed. Do not silently guess unverified personal dates.
When reporting anniversary spans, label where each date came from: (provided by you), (public date used: [label]), or (from My Events). Only attribute a date to My Events if I asked for them to be included and the tool actually returned it. Note that My Events are capped at 10 dates, so some of my saved events may not be analyzed.
For every gematria line, print the cipher as {canonical_name} ({code}). The name must match the map exactly. If a code is missing in the map, print the code only and mark it as unresolved instead of guessing.
From custom_date_analysis: for each custom date, show all anniversary spans (since_last_anniversary and until_next_anniversary) with their full duration variations (Months+Days, Weeks+Days, Days Only, etc.) and any gematria matches found for those span values.
From quick_insights.cross_context_patterns: for each value, build a card in this format — sort by most contexts first, flag 3+ contexts as highly significant:

**{value}** · {N} analyzed words • {N} database matches • {N} date calcs • {N} custom date calcs
Your Words: list each word + cipher + value from gematria_results that produced this value
Database Matches: list each word + cipher + value from word_database_matches where the matched value equals this number
Date Calculations: list matching date numerology entries from date_numerology
Custom Date Calculations: list matching span entries from custom_date_analysis (e.g. "1987-06-24 - Messi Born (Last Anniversary 68 Days) = 68")
List every number that appears in 2+ contexts. Sort by most contexts first — flag 3+ context patterns prominently as highly significant.
From quick_insights.high_confidence_matches: show words matching in 3+ ciphers with database entries.
From quick_insights.word_correlations: show pairs of analyzed words that share the same value in multiple ciphers.
From quick_insights.most_connected_words: list the words with the most total matches across all contexts.
From quick_insights.known_numbers_summary: list known numbers found in the analysis with their database descriptions.
Call headlines_decode and flag any headline values that also appear in cross_context_patterns.
Call date_text_gematria for the event date and report the written-date forms (e.g. "September Second", "September Second Twenty Twenty Six"). Flag any of those values that also appear in cross_context_patterns.
For the 1–3 strongest values only, call number_properties and note anything structural (prime, triangular, Fibonacci). Skip this if nothing stands out — do not run it on every value.
State the search space. Report how many phrase variants × ciphers were calculated, and how many total matches came back. Because a large search space produces coincidental matches, explicitly say when a 2-context pattern is within what random chance would produce. Do not present an expected result as remarkable.
Before finalizing, run a strict label check: every mentioned cipher label must be verifiably present in list_ciphers output and aligned to the exact code used in results. Never blend or shorten two cipher names into a new one.
End the report with a short legend listing each cipher code you used and its exact name from list_ciphers, so the labels can be checked.
Report only what tools returned for calculations and matches. Do not add connections, computed values, interpretations, or alternate cipher names of your own.
If nothing meaningful turned up, say so plainly rather than presenting weak matches as findings. Finish by offering /capture to save anything worth keeping.

/predict Find future dates that match a word's gematria, plus eclipse and span correlations

Never compute gematria or date math yourself. Every value and every date must come from GematriaLens tool results.
Call list_ciphers and build a canonical {code -> name} map. Label every cipher using that map exactly — never invent, blend, or shorten a cipher name.
Call predictive_calendar_analyze with the word or name I give you and the year I ask for (default: current year). Results come back as cipher_matches, keyed by cipher name, each holding a value and a list of matches with date, display, variants and is_direct.
First show the gematria values as a table: cipher name, code, value, and how many dates matched that value. Say explicitly if a value produced no matches.
Group results by date, not by cipher, and sort by the number of distinct ciphers matching that date. Every date line must show, for each cipher that hit it, the value and the numerology variant that produced the match — never the cipher name alone. Use this format:

**February 28** — Reverse Ordinal (ro) = 47 via Month + Day · Reverse Reduction (rr) = 29 via Day of Year

Take the value from the cipher’s value field and the variant text from that match’s variants list, exactly as returned. If a match has several variants, list them all.
Mark any match where is_direct is true — the calendar date itself spells the value. Always show the numeric date form next to it so the correspondence is visible rather than asserted, e.g.

**February 22** — 2/22 → 222 — Reverse Satanic (sar) = 222 (direct)
**May 28** — 5/28 → 528 — Sumerian (su) = 528 (direct)

The digits split by month and day according to the date_format used (US = M/D, EU = D/M) — state which format the run used. A direct mapping is a stronger correspondence than a derived variant like “Day of Year” or “Days Left in Year”, so list direct matches in their own group.
Do not report only the multi-cipher dates. The named variants — Direct Mapping, Day of Year, Days Left in Year, Month + Day, Day + Month and their Reversed forms — are meaningful even when a single cipher hits them, and must each be listed. Example: if 47 matches February 16 via Day of Year (Feb 16 is the 47th day of 2026), that belongs in the report. Variants written as arithmetic digit sums (e.g. (15) + (6) + (26)) are the weakest kind — group them separately or summarise their count rather than mixing them in with the named variants.
Structure the report as: (1) direct mappings, (2) named-variant matches such as Day of Year and Days Left in Year, (3) dates hit by 2+ ciphers, (4) a count of the remaining digit-sum matches. State the total so nothing looks hidden.
Only include dates the tool returned. Never add dates from your own knowledge, memory, or earlier conversations.
For the strongest 1–3 upcoming dates, call calculate_date_duration from today to that date. Report the duration variations and any eclipse data the tool returns.
If I give you a birth date or other significant date, analyse the spans too — predictive_calendar_analyze does not do this on its own. Call calculate_date_duration from that date to each of the top matched dates, and report the duration variations plus any value that equals one of the cipher values above. For anniversary spans around a single event date, use analyze_text with custom_dates instead. Only use dates I actually gave you.
Call list_upcoming_calendar_events and note any of my saved calendar events that fall on or near a matched date.
State the search space. Say how many ciphers were analyzed and how many dates matched out of the days in the year. With enough ciphers some matches are expected by chance — say so plainly instead of implying every hit is meaningful.
These are numerological correspondences, not forecasts. Do not describe a matched date as something that will happen.
End with a cipher legend, and offer save_to_calendar_id or /capture if I want to keep the matches.

/capture Save and organize research findings into your databases, events, and decode cards

Only save words and dates that were explicitly found using GematriaLens tools earlier in this conversation — do not invent or add your own.
For any saved card line with a cipher, preserve the exact canonical label from tool output (or from list_ciphers code->name mapping). Never rename, paraphrase, or infer cipher names.
If I already named a database earlier in the conversation, use it. Otherwise ask me which database to use — if I say I don’t have one, offer to create it with create_word_database and confirm before doing so.
For each word to save, call database_add_word with the exact word as it appeared in results.
For any dates discussed, call create_my_event with a descriptive label I provide or confirm with me first.
Call create_card to create a Decode Card titled with the topic. Only include values that came from tool results.
List exactly what was saved: words added (with database name), events saved, and the card share link.

Skills persist across all conversations — save once, invoke with a slash command anytime. Requires an AI client with the GematriaLens connector added.

Questions about limits, skills or unexpected AI answers? See the AI Agent FAQ.

Compatible AI Clients

Works with any client supporting the Model Context Protocol (MCP) standard.

Claude.ai Any plan • No API key needed • OAuth connect

Claude Desktop Free tier available • Requires Node.js • API key

GitHub Copilot VS Code agent mode

Claude Code Terminal • Remote or local

Cursor Built-in MCP support

ChatGPT Connectors • Availability varies by plan

Goose Open source

Any MCP Client Streamable HTTP transport

The https://gematrialens.com/mcp endpoint accepts either OAuth (authorize in the browser, no key to manage) or an API key sent as a Bearer token — use whichever your client supports. Claude.ai is the easiest: add the connector URL and authorize with your account. Claude Desktop expects a local command rather than a URL, so it reaches the same remote endpoint through the mcp-remote bridge, which needs Node.js (v18+) for npx.

Ready to Automate Your Research?

Create a free account to get started with the AI Agent.

Create Free Account Log In