Connects AI clients to VK for profile, community, wall, photo, feed, likes, and statistics operations.
Copy the AI prompt to install this server into Claude Code, Cursor, or another agent — or use 1-click editor setup below.
This server is confirmed live — we successfully called its tools/list endpoint directly (see the verified badge above). We haven't yet sandbox-tested the stdio install command below specifically, which is a separate, ongoing check.
💡 Paste the JSON block into your client's configuration file under mcpServers, then restart the application.
Inspect callable tools, capabilities, and parameters exposed to AI agents by Vk MCP Server.
vk_users_getLook up VK users by numeric ID or short name (e.g. durov). Use this to resolve a name to an ID before calling other tools, or to check whether a profile is closed.
vk_users_searchFind VK users by name, optionally narrowed by city, country, sex or age. Returns matches with a total count; page through them with offset.
vk_wall_getRead posts from a user or community wall, newest first. Pass domain for a short address (durov) or owner_id for a numeric one — negative for a community, positive for a person. Each post carries its likes, reposts, comments and views.
vk_wall_postPublish a post on a wall. To post in a community, set owner_id to the community ID with a minus sign (-123) and from_group true, otherwise it appears as your personal post on the community wall. Attach media with the attachment string that vk_photos_upload_wall returns. Returns the new post_id.
vk_wall_create_commentAdd a comment to a post. owner_id is negative for a community. Returns the new comment_id.
vk_wall_get_by_idFetch particular posts by their full IDs in {owner_id}_{post_id} form, e.g. -1_340393. Use it to re-read a post you already know about, such as one you just published.
The bulatko/vk-mcp-server MCP server connects an MCP-compatible client to VKontakte’s API. It provides 19 tools covering user lookup and search, wall reading and publishing, comments, post editing and deletion, community discovery and membership, friend lists, the token owner’s newsfeed, photo uploads and album browsing, likes, and community statistics.
The read tools can resolve users or communities by numeric ID or short name, inspect profile or community details, search by filters, and retrieve paginated results. Wall responses include engagement counters such as likes, reposts, comments, and views. Write operations include publishing posts, adding comments, editing posts, deleting posts, and uploading photos for later attachment.
The server translates MCP tool calls into VK API requests and returns structured, typed results. List operations report match counts and continuation offsets, allowing a client to request additional pages instead of treating the first response as complete. Prompts are also included for workflows such as community digests, engagement reports, audience snapshots, and community searches.
VK access depends on the token type and the account or community behind it. User tokens are needed for a user’s communities, friends, and personal newsfeed. Community statistics require the token owner to administer the target community. Posting, editing, and photo uploads require a community token with the relevant permissions. Public profile, wall, and community reads can use a service token or a token obtained through the VK ID login flow, but that login token does not provide the broader operations.
Hosts that support MCP Apps can render walls, communities, and profiles as cards. Other MCP clients still receive the same structured tool data without those visual renderings.
Install and run the package with npx -y vk-mcp-server. MCP client configuration supplies the access token through VK_ACCESS_TOKEN. The server can start and expose its tools without a token, but tool calls require one. The repository documents configurations for Claude Desktop and Claude Code, and also provides a one-click installation path for VS Code.
VK_TIMEOUT_MS controls the request timeout and defaults to 30,000 milliseconds. VK_API_BASE can replace the default VK API endpoint with a mirror or proxy. The --check option identifies the token type and tests which operations it can reach without invoking write methods. The --login <APP_ID> option starts the VK ID sign-in flow for a reading token.
VK’s permissions are not uniform across token types. A service token cannot call user methods, private profiles may deny access, and communities can hide their member lists. Friend retrieval only works when the target profile exposes its friends. Statistics are restricted to community administrators.
User tokens are rate-limited to a few requests per second. The server retries rate-limit responses with backoff up to three times, and it reports timeout, captcha, HTTP, and VK API errors with explanatory messages. Deleting a wall post is irreversible, while editing replaces the existing text or attachments rather than appending to them.
The bulatko/vk-mcp-server MCP server is therefore best suited to VK workflows where the required token permissions are known in advance. Use numeric negative owner IDs for community walls, positive IDs for personal walls, and positive group IDs when uploading community photos.
Factual signals from GitHub, npm, and our automated checks — not a rating.
No reviews yet — be the first to share how this listing worked for you.
Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.
[](https://allmcps.com/mcp/bulatko-vk-mcp-server)<a href="https://allmcps.com/mcp/bulatko-vk-mcp-server"><img src="https://allmcps.com/api/badge/bulatko-vk-mcp-server?style=directory" alt="Vk MCP Server on AllMCPs" /></a>