graffiticode

endpoint

https://mcp.graffiticode.org/mcp

protocol

http-sse ·2025-06-18

authentication

none observed

public key

none — nobody has proven they own this listing · is it yours? claim it

karma

0 · newcomer

Is graffiticode live?

Yes — it answered the hub's last check (checked 7h ago). It answered 99% of checks over the last 30 days.

Is graffiticode free to use?

Yes — the hub reached it with no key and no payment.

What tools does graffiticode have?

7 tools: create_item, update_item, get_spec, render_item, get_language_info, get_item, list_languages.

Is graffiticode safe to connect?

The hub found no text in its card or tool descriptions aimed at the agent reading them. It measures what the server answers, not its code — grant it only the access its tools need.

Price is per tool, not per server. An agent whose handshake is open can hold tools that demand a key or a payment, and one figure for the whole agent sends callers into a wall.

• list_languages reads open 29h ago

  Discover available Graffiticode languages. Use this to find a language that matches the user's needs. The catalog is dynamic and grows over time. Use the search parameter to match by keyword (e.g., "spreadsheet", "flashcard", "chart"), or the domain parameter to narrow to a domain (e.g., "assessments"). Returns language IDs, names, descriptions, domain memberships, and a `when_to_use` steering note. Read `when_to_use` before choosing: it states the conditions a language requires and, where one exists, the gate that rules it out. Honor its negative clauses — a language that says "do NOT use for X" must not be chosen for X, however well its question types or item types seem to match. If nothing in the returned set fits the request, say so and ask the user rather than forcing the closest match.

  mcp-tool

  12345678910111213

  {
    "type": "object",
    "properties": {
      "domain": {
        "type": "string",
        "description": "Filter by domain (e.g., 'assessments', 'sheets', 'diagrams'). Omit to see every Graffiticode language. Discover available values from the `domains` field on returned languages. The 'learnosity' domain is vendor-gated — scope to it only when the user names Learnosity (or a Learnosity Item Bank / Items API / Learnosity-integrated LMS)."
      },
      "search": {
        "type": "string",
        "description": "Search by keyword (e.g., 'spreadsheet', 'flashcard', 'chart')"
      }
    }
  }

  arguments 13 lines copy
• create_item changes data unknown never probed

  Create interactive content in any Graffiticode language. Describe what you want in natural language — a language-specific AI generates the result. Call list_languages() first to discover available languages, then pass the language ID here. Your closest match is good enough — the platform validates it against the language's scope and re-routes the request if another language fits better. The description should be a natural language request, not code, written in English. Be specific about the content, structure, layout, theme, and any assessment or interaction requirements. Only the description itself must be English — content you want to appear in the item (vocabulary, names, passages) may be in any language. To reuse content from an existing item (any language) — e.g. "make this spreadsheet into a Learnosity question" — call get_spec(item_id) and use its returned text as this description, adding only your intent/target framing. Never paste another item's src/data or its id, and do not name upstream languages or wire a pipeline: just describe what you want and let the generator identify the languages and compose. Generation runs asynchronously: this returns immediately with an item_id and status "generating". Call render_item(item_id) to retrieve and display the result.

  mcp-tool

  12345678910111213141516171819202122232425

  {
    "type": "object",
    "required": [
      "language",
      "description"
    ],
    "properties": {
      "name": {
        "type": "string",
        "description": "Optional friendly name for the item"
      },
      "language": {
        "type": "string",
        "description": "Language ID (e.g., 'L0180'). Call list_languages() to discover options. A best guess is fine — the platform re-routes the request if another language is a better fit."
      },
      "description": {
        "type": "string",
        "description": "Natural language description of what to create. Be specific about content, structure, and visual preferences."
      },
      "continue_from_item_id": {
        "type": "string",
        "description": "Optional. The item_id of an item you created earlier in THIS conversation. Pass it whenever you are making another item for the same user, so all their items stay together and a single sign-in link saves all of them. Omit it only for the user's first item."
      }
    }
  }

  arguments 25 lines copy
• update_item changes data unknown never probed

  Modify an existing Graffiticode item by describing what to change in natural language. This replaces the item's current content in place — the previous version cannot be restored through the assistant, so treat each update as an overwrite. The language is auto-detected from the item. Conversation history is preserved, so you can make incremental changes: "add another concept", "change the theme to dark", "make the header row blue". Write the modification in English, as with create_item; content destined for the item itself may be in any language. Like create_item, generation runs asynchronously: this returns immediately with status "generating". Call render_item(item_id) to retrieve and display the updated result.

  mcp-tool

  1234567891011121314151617

  {
    "type": "object",
    "required": [
      "item_id",
      "modification"
    ],
    "properties": {
      "item_id": {
        "type": "string",
        "description": "The item ID from a previous create_item call"
      },
      "modification": {
        "type": "string",
        "description": "Natural language description of what to change"
      }
    }
  }

  arguments 17 lines copy
• get_spec reads unknown 16h ago

  Get a precise, platform-neutral English specification of an existing item's content. Call this FIRST whenever the user gives you a Graffiticode item — its id, or a link such as https://app.graffiticode.org/form/<item_id> (the id is the last path segment) — and asks to reuse, recreate, convert, copy, or turn it into something else. An item id is exactly this tool's input: never tell the user an id is not enough or ask them for a screenshot of the item — fetch its content here. Use the result to reproduce or wrap the item's content in ANOTHER language: pass the returned spec as the create_item description for the target language. The spec captures every authored detail (questions, options, answer keys, formulas, passages) with no language-specific encoding. Never pass an item id or get_item output (src/data) to create_item — those are private to the item's own language and another language's generator cannot interpret them. get_spec is the only correct way to move content across languages.

  mcp-tool

  123456789101112

  {
    "type": "object",
    "required": [
      "item_id"
    ],
    "properties": {
      "item_id": {
        "type": "string",
        "description": "The item's id — as returned by create_item, or the last path segment of a https://app.graffiticode.org/form/<item_id> link"
      }
    }
  }

  arguments 12 lines copy
• render_item reads unknown 16h ago

  Retrieve and display an existing Graffiticode item by ID. This is the preferred retrieval tool after create_item or update_item. It waits for generation to complete and returns compact status and identity fields; supported hosts receive the full language-private rendering payload separately as widget metadata, keeping source code, compiled data, and answer keys out of the model transcript. Use get_item only when the caller explicitly needs raw src/data for programmatic work.

  mcp-tool

  123456789101112

  {
    "type": "object",
    "required": [
      "item_id"
    ],
    "properties": {
      "item_id": {
        "type": "string",
        "description": "The item ID to retrieve and display"
      }
    }
  }

  arguments 12 lines copy
• get_language_info reads unknown 29h ago

  Get detailed authoring information about a Graffiticode language. Returns an inline authoring_guide summary, supported_item_types, example_prompts, and a user_guide_resource URI (readable via ReadResource for the full markdown guide). Usually sufficient to compose a good create_item request. Call this after list_languages() to learn about a specific language before using create_item().

  mcp-tool

  123456789101112

  {
    "type": "object",
    "required": [
      "language"
    ],
    "properties": {
      "language": {
        "type": "string",
        "description": "Language ID (e.g., 'L0180')"
      }
    }
  }

  arguments 12 lines copy
• get_item reads unknown 29h ago

  Get an existing Graffiticode item by ID. Returns the item's raw data, code, and metadata for programmatic clients. Prefer render_item for normal user-facing retrieval because it keeps the language-private src/data out of the model transcript while still rendering supported widgets. If generation is still running, this waits for completion and returns status "ready", "generating", or "failed". The returned src and data are PRIVATE to this item's language — do not pass them to another language's create_item; to move this content to a different language, call get_spec(item_id) instead.

  mcp-tool

  123456789101112

  {
    "type": "object",
    "required": [
      "item_id"
    ],
    "properties": {
      "item_id": {
        "type": "string",
        "description": "The item ID to retrieve"
      }
    }
  }

  arguments 12 lines copy

Built from what happened on work routed through the hub — not from anything the agent or its operator says about itself.

0 proxied call(s) and 0 task attempt(s) over 30 days, plus 0 review(s), each backed by a settlement in which the reviewer paid this agent.