Skip to content

For developers

Versemora's data — traditions and their postulates with their support in the texts, the timeline, influences and disputes, the history of texts and the library — for AI assistants through an MCP server and for programs through a REST API. Free, no key, read-only.

MCP server

The Model Context Protocol is an open protocol through which assistants (Claude, ChatGPT, Cursor, VS Code and others) call external tools. Add the URL and the assistant can search traditions, read postulates with their places, the timeline and the library's texts, with links to the site's pages.

Server URL
https://versemora.xyz/mcp
  • Transport: Streamable HTTP, stateless; clients of protocol revisions 2025-03-26 … 2025-11-25 and 2026-07-28 are served.
  • Authentication: none. Every tool only reads public data (readOnlyHint).
  • Answer languages: Russian and English (the lang argument); every entity comes with a link to its page.
  • Postulate grades show support in the tradition's own texts, not whether a teaching is true.

A short description for models: /llms.txt

Connect

  • Claude — claude.ai and Claude Desktop

    1. Open Settings → Connectors (in newer versions Customize → Connectors).
    2. Click Add → Add custom connector.
    3. Name: Versemora. URL: https://versemora.xyz/mcp. No sign-in is needed — click Add.
    4. In a chat, turn Versemora on in the tools menu (+ → connectors).
    If your Claude Desktop has no connectors: claude_desktop_config.json (the mcp-remote bridge, needs Node.js)
    {
      "mcpServers": {
        "versemora": {
          "command": "npx",
          "args": ["-y", "mcp-remote", "https://versemora.xyz/mcp"]
        }
      }
    }

    Pro, Max, Team and Enterprise; the Free plan allows one custom connector. On Team and Enterprise an Owner adds it (Organization settings → Connectors). Connectors of claude.ai also appear in Claude Desktop.

  • Claude Code

    1. Run the command below in a terminal (add --scope user for all projects).
    2. Check with claude mcp list, or /mcp inside Claude Code.
    Terminal
    claude mcp add --transport http versemora https://versemora.xyz/mcp
    Or .mcp.json at the project root
    {
      "mcpServers": {
        "versemora": {
          "type": "http",
          "url": "https://versemora.xyz/mcp"
        }
      }
    }
  • ChatGPT

    1. Open chatgpt.com → Plugins (chatgpt.com/plugins), click + → Add custom MCP server.
    2. Name: Versemora. Server URL: https://versemora.xyz/mcp. Authentication: No authentication.
    3. Accept the warning (I understand and want to continue) → Create as a plugin, then install it and pick it in a chat.

    Web only; availability depends on your plan and workspace policy. In the older interface: Settings → Apps & Connectors → Advanced → Developer mode, then Create.

  • Cursor

    1. Add the server to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project), or via Settings → MCP → Add new server.
    2. Turn it on in the list of MCP servers.
    mcp.json
    {
      "mcpServers": {
        "versemora": {
          "url": "https://versemora.xyz/mcp"
        }
      }
    }
  • VS Code (GitHub Copilot)

    1. Command Palette → MCP: Add Server → HTTP → the server URL → name it versemora.
    2. Or add it to .mcp.json at the workspace root (or ~/.copilot/mcp-config.json).
    .mcp.json
    {
      "mcpServers": {
        "versemora": {
          "type": "http",
          "url": "https://versemora.xyz/mcp",
          "tools": ["*"]
        }
      }
    }
    The older .vscode/mcp.json format
    {
      "servers": {
        "versemora": {
          "type": "http",
          "url": "https://versemora.xyz/mcp"
        }
      }
    }
  • Other clients and testing

    1. Any MCP client with the Streamable HTTP transport: the server URL, no authentication.
    2. Test the server with the MCP Inspector:
    Terminal
    npx -y @modelcontextprotocol/inspector --cli https://versemora.xyz/mcp --transport http --method tools/list

Tools

22

Every tool only reads. Lists page with offset and limit; the ids one tool gives work in the others.

  • searchSite search: traditions, events, people, lineages, editions.
  • search_textsFull-text search in the library's editions.
  • list_religionsReligions and their direction trees, with postulate counts by grade.
  • get_traditionA religion or direction: dates, texts, postulates with grades, branches, editions.
  • get_postulateA postulate: its grade and why, places for and against, inherited places, the tradition's own words.
  • get_timelineThe timeline: directions, events and world history, by scope and years.
  • get_eventAn event: date, people, texts, what led to it and what came of it.
  • get_personA person: dates, biography, works, versions of texts, events.
  • list_lineagesLineages of texts: from manuscripts to editions and translations.
  • get_lineageOne lineage graph: versions, the links between them, events.
  • list_influencesInfluences: who took what from whom, when, and how firmly it is established.
  • get_influenceOne influence: evidence, strength, the state of scholarship.
  • list_claimsClaims of traditions about each other and themselves, with the evidence's verdict.
  • get_claimOne claim: where it is made, the evidence, the verdict and why.
  • list_conflictsConflicts: statements of texts that other sources contradict.
  • get_conflictOne conflict: the statement, what contradicts it, the tradition's reply.
  • list_contestedContested texts: what a tradition relies on that draws objection.
  • get_contestedOne contested text: what it says, the objection, the tradition's reading, critics.
  • list_editionsThe library's editions with language, date, role and licence.
  • get_editionAn edition: source, licence and attribution, contents with page ids.
  • read_passageThe text of a place: a verse, a range or a chapter; protected editions only a short excerpt.
  • compare_translationsOne place in other editions and translations, by recorded correspondences.

Limits

  • Up to 120 requests a minute per IP address (a burst of as many at once); over it, 429 with Retry-After. Every answer carries RateLimit-Limit, RateLimit-Remaining and RateLimit-Reset.
  • Protected editions: at most 3 units and 400 characters per answer, and at most 40 excerpts per 10 minutes per address.
  • A tool answer is at most ~40,000 characters; long lists and passages come in parts (offset/limit, next_offset, continue_with). Request bodies up to 64 KB.
  • The limits are kept in each server instance's memory: a bound on abuse, not an exact quota. No key, no sign-up.

REST API

The same data server directly: JSON, GET only, no key. Texts come in both languages as {"ru", "en"}; directions, postulates and readings take ?lang=ru|en. Answers are cacheable (cache-control: public, max-age=300, ETag and 304). Errors are {"error": {"code", "message"}}: 404 for an unknown id, 503 index_building with Retry-After while the server builds its index after a restart.

Base URL
https://versemora.xyz/v1
Example
curl -s 'https://versemora.xyz/v1/traditions/lutheranism/postulates/scripture-sole-rule?lang=en'

Traditions

  • /v1/traditions — religions and direction trees, the grade legend
  • /v1/traditions/{id}?lang= — a religion or direction: texts, postulates with grades, branches
  • /v1/traditions/{direction}/postulates/{postulate}?lang= — a postulate: grade, places for and against, the tradition's words
  • /v1/traditions/timeline?religion=&direction=&history=1 — lanes, events, world history

Relations and critique

  • /v1/relations/influences, /v1/relations/influences/{id}
  • /v1/relations/claims, /v1/relations/claims/{id} — claims and verdicts
  • /v1/critique/conflicts[/{id}], /v1/critique/contested[/{id}], /v1/critique/basis/{direction}, /v1/critique/phases/{religion}

History of texts

  • /v1/history/people[/{id}], /v1/history/events/{id}
  • /v1/history/lineages[/{id}], /v1/history/editions/{id}
  • /v1/media[/{id}] — photographs and portraits with author and licence

Library

  • /v1/library/catalog, /v1/library/editions[/{id}], /v1/library/editions/{id}/pages/{page}
  • /v1/library/passage?edition=&ref= — a unit, range or chapter (up to 200 units)
  • /v1/library/resolve?q= — 'Matt 22:39' → canonical refs
  • /v1/library/search?q= — full-text search
  • /v1/library/parallel?ref=&from=, /v1/library/editions/{id}/readings?place=&lang= — the same place in other editions

Cache the answers and do not download the library whole: the server is shared.

Licences and attribution

  • Cite Versemora's data (traditions, grades, relations, verdicts, the history of texts) with a link to the page on the site — every entity has one (url).
  • Texts belong to their rights holders. Every edition has its source, licence and attribution line (license.attribution, license.url); give it with a quotation.
  • Protected editions (licence class research_use, “© online only”) are read on the site; the MCP server gives only a short excerpt of one (up to 3 units, 400 characters) with its usage notice. Do not redistribute them, assemble them whole or republish them.
  • Photographs of manuscripts and finds come from open sources; each names its author and licence — keep both.
  • The site's terms of use apply here too: do not try to get around the limits of protected editions, and do not download texts in bulk.

Terms of use →