> ## Documentation Index > Fetch the complete documentation index at: https://docs.zuddl.com/llms.txt > Use this file to discover all available pages before exploring further. # Upload Speaker Picture > **Tool name:** `v1_upload_speaker_pic` · **Behaviour:** ✏️ Write Upload a profile picture for a speaker, replacing the current one. The image must be JPEG or PNG, at most 4 MB, and at least 80x80 px (300x300 recommended); its pixel data is fully decoded and anything incomplete or corrupt is rejected with IMAGE_CORRUPT. Provide exactly ONE source, in this order of preference: (1) imageFile — for an image the user attached in this conversation. Hosts that support file inputs fill this in themselves; you cannot construct it. (2) imageUrl — a publicly reachable https URL of the image. Redirects are not followed, and invalid, private or unresolvable URLs are rejected (INVALID_URL / PRIVATE_ADDRESS / UNRESOLVABLE). (3) imageBase64 — ONLY when you can run a real encoder over a real file on disk (e.g. `base64 -i photo.png`); send the output verbatim, uncompressed, unresized, not re-encoded and not truncated. NEVER hand-write, reconstruct, guess or abbreviate base64, and never try to encode an image you were merely shown in the conversation — you do not have its bytes and the result will be rejected. If the user attached an image and imageFile is unavailable, ask them for a URL instead. Returns the new picUrl. **Same as Public API:** [Upload a speaker's profile picture](/api-reference/speakers/upload-a-speakers-profile-picture) (`POST /v1/events/{eventId}/speakers/{speakerId}/pic`) ## OpenAPI ````yaml https://api.prod.zuddl.com/api/v3/api-docs/mcp-tools post /tools/v1_upload_speaker_pic openapi: 3.1.0 info: title: Zuddl MCP Tools Reference description: >- The Zuddl MCP server lets AI assistants such as Claude and ChatGPT work with your Zuddl organization through the [Model Context Protocol](https://modelcontextprotocol.io). This reference lists every tool the server offers. It is generated from the live server, so it always matches what your assistant sees. # Connecting Add `https://setup.zuddl.com/api/mcp` as a remote MCP server in your assistant. It signs you in to Zuddl with OAuth and then calls tools on your behalf, with your permissions. AI features must be enabled for your organization. # Calling a tool Assistants call a tool with a JSON-RPC `tools/call` request whose `name` is the tool name and whose `arguments` match the request body documented on the tool's page. The `/tools/` paths in this reference identify tools; they are not HTTP endpoints. # Behaviour labels | Label | Meaning | | --- | --- | | 🔒 Read-only | Reads data and changes nothing. | | ✏️ Write | Creates or updates data, but does not delete it. | | ⚠️ Destructive | Deletes or removes data. Most assistants ask you to approve these calls. | | 🔁 Idempotent | Calling it again with the same arguments has no further effect. | A tool that mirrors a Public API endpoint links to that endpoint's page. MCP-only tools have no Public API equivalent. version: 1.0.0 servers: - url: https://setup.zuddl.com/api/mcp description: Zuddl MCP server security: - oauth: [] tags: - name: Events - name: Attendees - name: Sessions - name: Speakers - name: Sponsors - name: Locations - name: Teams - name: Brand Kits - name: Registration flows & tickets - name: Analytics paths: /tools/v1_upload_speaker_pic: post: tags: - Speakers summary: Upload Speaker Picture description: >- **Tool name:** `v1_upload_speaker_pic` · **Behaviour:** ✏️ Write Upload a profile picture for a speaker, replacing the current one. The image must be JPEG or PNG, at most 4 MB, and at least 80x80 px (300x300 recommended); its pixel data is fully decoded and anything incomplete or corrupt is rejected with IMAGE_CORRUPT. Provide exactly ONE source, in this order of preference: (1) imageFile — for an image the user attached in this conversation. Hosts that support file inputs fill this in themselves; you cannot construct it. (2) imageUrl — a publicly reachable https URL of the image. Redirects are not followed, and invalid, private or unresolvable URLs are rejected (INVALID_URL / PRIVATE_ADDRESS / UNRESOLVABLE). (3) imageBase64 — ONLY when you can run a real encoder over a real file on disk (e.g. `base64 -i photo.png`); send the output verbatim, uncompressed, unresized, not re-encoded and not truncated. NEVER hand-write, reconstruct, guess or abbreviate base64, and never try to encode an image you were merely shown in the conversation — you do not have its bytes and the result will be rejected. If the user attached an image and imageFile is unavailable, ask them for a URL instead. Returns the new picUrl. **Same as Public API:** [Upload a speaker's profile picture](/api-reference/speakers/upload-a-speakers-profile-picture) (`POST /v1/events/{eventId}/speakers/{speakerId}/pic`) operationId: v1_upload_speaker_pic requestBody: description: The tool's `arguments`, as sent in a JSON-RPC `tools/call` request. content: application/json: schema: type: object properties: eventId: type: string description: Unique identifier of the event speakerId: type: string description: Unique identifier of the speaker imageFile: type: object description: >- An image the user attached in this conversation. The host fills this in for file inputs — do not construct or invent it. Preferred source when available. properties: download_url: type: string description: Short-lived URL the server downloads the image from file_id: type: string description: Host's identifier for the uploaded file file_name: type: string description: Original file name, if the host sends one mime_type: type: string description: >- Declared MIME type; advisory only, the format is sniffed from the bytes imageUrl: type: string description: >- Publicly reachable https URL of the image (JPEG/PNG, max 4 MB, min 80x80 px). imageBase64: type: string description: >- Base64 of the image FILE's raw bytes, produced by an actual encoder (e.g. `base64 -i photo.png`) and pasted verbatim — uncompressed, unresized, not re-encoded, not truncated. A data: URI prefix is allowed. Only usable when you can read a real file; never for an image you were only shown in the conversation. required: - eventId - speakerId required: true responses: '200': description: A JSON-RPC `tools/call` result. The tool's output is in `content`. components: securitySchemes: oauth: type: http description: >- OAuth 2.1 access token. MCP clients obtain it themselves (dynamic client registration + PKCE) when you connect them to the server URL above. scheme: bearer ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.