MCP Servers for Social Media Automation: 2026 Statistics
Of 585 social media MCP servers in the MCP Registry or with 5+ GitHub stars that have a dated public repo, 364 had a commit in 90 days.
Platforms: X, LinkedIn, Facebook, Instagram, Threads, TikTok, YouTube, Pinterest, Bluesky and Reddit. Deduplicated and checked 12 October 2026. 306 servers have no public repo we could date, Blotato's hosted server among them.
How we did it: we read every official MCP Registry entry and every GitHub repository with 5 or more stars that matched our social media search, then labeled each server from its own README, manifest, registry entry or docs, with the quote behind every label except those that read not stated. AI agents classified each server, and a separate set of AI agents re-checked every label against the live repository or docs page (the registry snapshot for servers listed only in the registry). Disagreements were settled under written rules. We never called any MCP server's tools. Blotato runs a vendor MCP server and is counted under the same rules. Method.
How many social media MCP servers are there?
We counted 891, deduplicated across the official MCP Registry and GitHub. Every one acts on a named platform's accounts or content. Servers that only download videos, fetch transcripts or run ads are counted separately (115 download or transcript servers and 72 ads-only servers, not included above). YouTube, X and LinkedIn have the most servers, and Pinterest has the fewest.
Table view
| Platform | Servers | Vendor | Independent | Platform | Can publish |
|---|---|---|---|---|---|
| YouTube | 359 | 243 | 116 | 0 | 147 |
| X | 344 | 230 | 113 | 1 | 214 |
| 330 | 253 | 77 | 0 | 182 | |
| 314 | 253 | 61 | 0 | 175 | |
| TikTok | 272 | 230 | 42 | 0 | 147 |
| 193 | 153 | 40 | 0 | 138 | |
| 160 | 97 | 63 | 0 | 40 | |
| Threads | 155 | 134 | 21 | 0 | 126 |
| Bluesky | 154 | 121 | 33 | 0 | 121 |
| 128 | 110 | 18 | 0 | 103 |
Source: Blotato, MCP Servers for Social Media Automation: 2026 Statistics, checked 12 October 2026. A server that acts on several platforms is counted once in the total and once under each platform. Blotato, which publishes this study, is one of the vendor servers on 9 of these platforms. Download PNGLink to chart
Which social media platforms have an official MCP server?
X is the only one of the 10 whose own GitHub organization, on the pages we checked, ships an MCP server with a tool that publishes posts, and that repository's last commit before our check was on 9 April 2026. Meta and TikTok build MCP servers for their ads platforms, and Meta also builds ones for developer apps and WhatsApp. Pinterest has announced one. On the pages we checked, LinkedIn, YouTube, Bluesky and Threads document none. So for organic posting on nine of the ten, the MCP options in this count are vendor and independent servers.
| Platform | Owner-built MCP server | What its own pages say | Source |
|---|---|---|---|
| X | Publishes posts (self-run repo) | X's own GitHub organization publishes xmcp, a server you run yourself, whose tool list includes createPosts. X also documents a hosted server whose capability table lists Articles and bookmarks as its writes, and a separate Ads MCP server. | github.comgithub.com |
| Ads and developer servers, no post tool | Meta documents remote MCP servers for Meta Ads, for managing developer apps, and for WhatsApp Business. None of their documented tools creates an organic Facebook post. | developers.facebook.com | |
| Ads server, no post tool | Meta's Ads MCP server can promote an existing Instagram post. None of Meta's documented MCP tools creates an organic Instagram post or reel. | developers.facebook.com | |
| TikTok | Ads server, no post tool | TikTok documents the TikTok for Business MCP Server for its ads platform. Its documented tool list has no tool for publishing an organic TikTok post. | business-api.tiktok.combusiness-api.tiktok.com |
| Announced, not launched | Pinterest's developer site announces a Pinterest MCP Server and says it has not launched yet, with a sign-up for launch news. | developers.pinterest.com | |
| Developer tool, no post tool | Reddit's GitHub organization publishes devvit-mcp, a server for writing apps on Reddit's developer platform. It does not act on Reddit accounts or posts, so it is not counted as a social media MCP server. | github.com | |
| None found | No LinkedIn MCP server is documented on the pages checked: LinkedIn's developer documentation on Microsoft Learn and LinkedIn's GitHub organization. | Pages listed in the Method | |
| YouTube | None found | No YouTube MCP server is documented on the pages checked: Google Cloud's list of supported MCP products, Google's MCP repository, the Workspace MCP guide, the YouTube Data API docs and Google's GitHub organizations. | Pages listed in the Method |
| Bluesky | None found | No Bluesky MCP server is documented on the pages checked: docs.bsky.app, atproto.com and Bluesky's GitHub organization. | Pages listed in the Method |
| Threads | None found | No Threads MCP server is documented on the pages checked: the Threads API documentation index and Meta's MCP hub. | Pages listed in the Method |
How do social media MCP servers sign in to your account?
344 of the 891 servers can publish posts. 193 sign in through a vendor: you connect your accounts in the vendor's app and the server calls the vendor, which holds the connection. 98 use your own credentials on the platform's API, such as a developer app you register or a Bluesky app password. 41 default to your cookies, your password, browser automation or a private-API library. 57 in all offer one of those routes somewhere in their setup, those 41 included. Where a server is listed changes the answer. Among servers listed only in the MCP Registry, 162 of 200 go through a vendor. Among servers found only on GitHub, 75 of 119 use your own credentials and 19 a vendor. And of the 193 vendor-route servers, 138 do not say, on the pages we read, how the vendor reaches the platform.
Table view
| Default sign-in | All | Registry only | GitHub only | Both | Vendor servers | Independent servers | Platform |
|---|---|---|---|---|---|---|---|
| Through a vendor that holds your connection | 193 | 162 | 19 | 12 | 175 | 18 | 0 |
| Your own credentials on the platform API | 98 | 17 | 75 | 6 | 4 | 93 | 1 |
| Unofficial: cookies, password, browser or private API | 41 | 11 | 23 | 7 | 13 | 28 | 0 |
| Not stated in its files | 12 | 10 | 2 | 0 | 9 | 3 | 0 |
Source: Blotato, MCP Servers for Social Media Automation: 2026 Statistics, checked 12 October 2026. Sign-in is the route a server's publish tool needs, as its own files describe it. Blotato, which publishes this study, is listed only in the MCP Registry and is one of the 193 vendor-route servers. Download PNGLink to chart
Your own credentials
You register an app or create a token on the platform. The server calls the platform's official API with it.
A vendor
You connect your accounts once in the vendor's app. The server calls the vendor, and the vendor holds the platform connection.
A scraping service
A data service collects public posts and profiles without your platform login. The server calls that service.
Unofficial access
The server uses your cookies or password, drives a browser, or calls a private API the platform does not publish.
Are social media MCP servers maintained?
Of the 585 servers with a public repo we could date, 364 had a commit in the 90 days before 12 October 2026, 77 in the 91 to 180 days before, and 144 are older or archived. Vendor servers with a public repo had a recent commit more often than independent ones: 222 of 255 against 142 of 329. The same gap shows by listing: 218 of 264 servers listed only in the registry with a dated repo had a recent commit, against 96 of 263 found only on GitHub, so the vendor and independent comparison may partly reflect where servers are listed. A recent commit shows activity, not that a server works. 257 of the 512 vendor servers have no public repo we could date and are not in that comparison, Blotato's included. The one platform-built server, X's xmcp, last had a commit on 9 April 2026.
Table view
| Builder | 90 days or less | 91 to 180 days | Older or archived | No dated public repo |
|---|---|---|---|---|
| Vendor | 222 | 20 | 13 | 257 |
| Independent | 142 | 57 | 130 | 49 |
| Platform | 0 | 0 | 1 | 0 |
Source: Blotato, MCP Servers for Social Media Automation: 2026 Statistics. Last non-bot commit on the default branch before 12 October 2026, from GitHub. Bars show shares of each builder's servers with a dated public repo. Blotato, which publishes this study, has no public repo and is not in these bars. Download PNGLink to chart
Find a social media MCP server
Pick a platform to see its servers, up to 40 at a time, with who built each one, how it signs in and when it last had a commit, and the quote behind the sign-in label. The CSV download has every server.
Loading the server list.
What this means if your agent posts to social media
On nine of these ten platforms we found no owner-built server that publishes, so the choice is between a vendor and an independent server. Check two things before you connect one. First, how it signs in: a server built on the platform's API with your own app goes through the platform's own sign-in. A vendor server does too only when the vendor connects your account on the platform's authorization page, and 138 of the 193 vendor-route servers do not say, on the pages we read, how they reach the platform, so ask the vendor how it connects your account before you do. A server that uses your cookies or password logs in as you outside the platform's sign-in. Second, its last commit: 130 of the 329 independent servers we could date had no commit in the last 180 days or are archived. Related: what platform rules let an AI agent do, which platforms broke their APIs and the social media MCP servers we compared.
Where Blotato lands in this study
Blotato publishes this study and sells a hosted social media MCP server, counted here as one of the 512 vendor servers. Its facts below come from help.blotato.com, read on 12 October 2026.
- Blotato runs a hosted MCP server with no public repository, so its commit history cannot be measured the way this study measures servers with public repos. Blotato docs
- It signs in through Blotato: you connect each social account on the platform's own authorization page, and Blotato holds that connection. That is the same trust question any vendor server raises. Blotato docs
- It publishes to 9 of the 10 platforms in this study. Reddit is not one of them. Blotato docs
- Comments and direct messages work on Instagram and Facebook only. Blotato docs
- API and MCP access needs a paid plan. The free trial does not include it. Blotato docs
How has social media's share of new MCP repositories changed?
Added 13 October 2026, from GitHub, npm and search data collected on 12 and 13 October 2026. A separate test from the count above, with its own data and rules.
Social media's share of new MCP repositories on GitHub (keyword matches, not checked servers) stepped up once, in February and March 2026, and stayed between 1.48% and 1.72% every month from April through September 2026. Before collecting any data, we registered a test asking whether the share from April to September 2026 was at least 10% higher than from October 2025 to March 2026, after adjusting, from small samples, for the share of matches that are real servers. Because the step came inside that earlier period, the gap between the periods is smaller than the chart's jump. The share formula registered then, applied to these same samples, gives a 95% low end of 16.7%, above the 10% bar. About six minutes after collection began, before any figure was computed, we made the share gate stricter by also adjusting for the share of MCP repositories that contain MCP code. That revision also loosened the completeness rule, and a later change raised the result slightly, as the block below explains. The share went from 1.28% to 1.57%, a 23% increase before adjustment, and the stricter version's 95% low end came to 7.2%, so it failed. We had planned to publish only if the test passed. We are publishing it anyway, because dropping a test that failed would make the ones we publish harder to trust.
Table view
| Month created | Share matching social media terms |
|---|---|
| Oct 2025 | 0.94% |
| Nov 2025 | 1.05% |
| Dec 2025 | 1.01% |
| Jan 2026 | 0.98% |
| Feb 2026 | 1.45% |
| Mar 2026 | 1.64% |
| Apr 2026 | 1.57% |
| May 2026 | 1.54% |
| Jun 2026 | 1.48% |
| Jul 2026 | 1.54% |
| Aug 2026 | 1.72% |
| Sep 2026 | 1.54% |
Source: Blotato, MCP Servers for Social Media Automation: 2026 Statistics, growth section added 13 October 2026. Repositories by the month they were created, on GitHub, forks excluded. Keyword matches, counted once per owner per month with no star floor, before any check of whether a repository is a real server, so these shares are not counts of servers. Blotato, which publishes this study, sells a hosted social media MCP server with no public repository, so that server is in none of these figures. Download PNGLink to chart
How we tested this
| Gate | Question | Result | What we found | Bar |
|---|---|---|---|---|
| G0 | Enough social media matches to test | Pass | 1,223 keyword matches in the first period (owner-capped, not servers) | at least 30 |
| G1 | More social media matches in the later period | Pass | 2.35x, low-end 2.26x | low-end 1.25x or more |
| G2 | Match counts not past a peak | Pass | Jul to Sep 2026 was the highest quarter | latest quarter's count at or above the one before, and at least 80% of the earlier peak |
| G3 | More search and download demand | Pass | US searches 4.48x (6 terms, Mar to Aug 2026 against Sep 2025 to Feb 2026), npm downloads 1.21x (164 packages, the study periods) | searches 1.10x or downloads 1.25x, each with its largest term or package removed, and searches not below 0.80x |
| G4 | Social media share of new MCP repositories at least 10% higher in the later period | Fail | 1.28% to 1.57%, low-end increase 7.2%, latest quarter 1.60% against 1.53% | low-end increase 10% or more, and the latest quarter's share at least 90% of the one before |
| G5 | G1 and G4 hold without matches whose only match term is the largest group (YouTube), and without the top month (August 2026) | Fail | G4 low-end increase without YouTube-only matches 9.8%, without August 2026 5.0% | G1 and G4 still pass |
- Registration. The 10% bar and the plan to publish only if the test passed were set on 12 October 2026, before collection began. About six minutes after collection began, before any figure was computed, the share gate was made stricter: it also adjusts for the share of MCP repositories that contain MCP code. With the earlier formula, on the same samples, the 95% low end is 16.7% against the 10% bar. The same revision loosened the completeness rule from an exact match to a small tolerance. About ten minutes into collection, the share was limited to matches that also appear in the MCP list, which slightly raised the observed change and its low end. After collection, the completeness rule stopped the study and was changed to count missing repositories instead of whole windows. Each change was logged before any growth figure was computed, and we decided to publish after seeing the result.
- Missing data. The social media list matched GitHub's own counts. GitHub's MCP search did not return 21 later-period social media matches (counted before the owner cap). Counting every one of them in the way most favorable to a pass lifts the low end only to 7.9%.
- Limits. These are keyword matches read from each repository's current name, description and topics, by the month it was created. Only samples were labeled, by AI models, and the 95% range covers sampling only. The registration timestamps are in our own change log, and no outside service time-stamped them. Blotato's own hosted server has no public repository, so it is not counted anywhere in this test.
- Full method and data. Method and gate results, monthly series (CSV), gate table (CSV), the 500 labeled sample repositories with quotes (CSV), full gate output (JSON), the pre-registration as registered before collection and in its final amended form, its change log and the labeling rules (redacted copies: names, internal file paths and internal tool names removed), and the analysis scripts unchanged: analyze_trend.py, collect_trend.py, collect_npm_a2.py, recheck_cells.py, run_d_months.py, fetch_samples.py and study6_collect.py. Two short helper scripts, one that picks the spot-check sample and one that assembles the labels the gates use, are listed by hash in the change log and available on request.
Where the platforms' own pages disagree
This list covers disagreements in the platforms' own pages about their own MCP servers. Disagreements inside individual servers' files are noted in the dataset.
X: what its MCP server can write
X's hosted MCP server documents Articles and bookmarks as its writes, and its Posts row only fetches. The same docs say the hosted server reports itself as xmcp, and X's open-source xmcp repository lists createPosts. We count X as publishing on the strength of the repository, and report the hosted docs as they read.
TikTok: how many tools its MCP server loads
TikTok's help center says its full server loads around 400 tools. The tools table in TikTok's docs lists 377 tools mapped to its Business API plus one diagnosis agent. We publish no TikTok tool total.
Meta: how many official MCP servers it has
Meta's MCP hub table lists two servers, Meta Social Technologies and WhatsApp Business Tools. Its documentation index adds a third, the Meta Ads MCP server.
Method
Each server was labeled from its own files under rules written before collection, then amended during checking. The changes to what counts are listed below, and every clarification of how a rule applies is listed in the dataset download.
Full method
- What counts. An MCP server (code in the repository, or an entry in the official MCP Registry) with at least one tool that acts on X, LinkedIn, Facebook, Instagram, Threads, TikTok, YouTube, Pinterest, Bluesky or Reddit accounts or content. Nine are the platforms Blotato publishes to, plus Reddit. Excluded, with counts: no action on a named platform (352), not an MCP server (130), download or transcript only (115), ads only (72), Threads not Meta's (51), other (18), lists of servers (14), templates (10), forks (7), duplicates (7) and files that give no sign of what the server does (5). Servers that answer from a vendor's previously collected data are in no action on a named platform. A counted server whose files do not say whether it reads or writes is labeled capability not stated.
- Where we looked. Every entry in the official MCP Registry (a full pull of 42,427 entries on 12 October 2026) and GitHub repositories with 5 or more stars, not forks, found by platform-name and social media searches. Servers are deduplicated by repository, then by server address. 568 are only in the registry, 263 only on GitHub, and 60 in both. Other directories were not used: their terms bar bulk reuse, or they offer no open API.
- Labels. Builder: platform (the platform's own GitHub organization or developer site), vendor (a company whose own commercial product the server fronts) or independent. Can publish: a documented tool creates an organic post, so ads and replies do not count. Sign-in: for a server that publishes, the route its publish tool needs, and for any other server, the route its setup uses first. Maintained: not archived, with a commit by a person (not a bot) on the default branch within 90 days of 12 October 2026. Every builder, platform, capability and sign-in label, except labels that read not stated, carries a verbatim quote from the server's own files or registry entry, checked by script against the saved file. Commit dates come from GitHub's API.
- Changes made after collection, before counting. We widened what counts as Threads to include Threads named alongside other social platforms, because the original wording, which required Meta, the Threads API or threads.net to be named, missed servers that plainly act on Meta's Threads. We added servers whose listing says social media without naming a platform, when their own files name the platforms they act on. We found that gap while looking for Blotato's own registry entry, which describes itself only as social media. We also fixed search patterns that missed abbreviations such as FB and IG. Counting only servers our original search terms found, the count is 723, 293 of 485 had a recent commit, and of 222 publishing servers 91 sign in through a vendor and 85 with your own credentials. Our first registry pull also stopped early, so it was repeated in full. During checking we also clarified two inclusion rules: download and transcript servers count when a tool returns view counts, comments, search results or feed listings, and servers that answer from a vendor's previously collected data, rather than the platform at request time, are excluded. Every excluded candidate and its reason is in the dataset download.
- Sensitivity. Among servers whose GitHub repository has 10 or more stars (211), 100 had a recent commit, and of the 90 that publish, 50 use your own platform credentials, 16 a vendor and 23 an unofficial route by default. At 50 or more stars (70), 35 had a recent commit, and of the 29 that publish, 15 use your own credentials, 6 a vendor and 8 an unofficial route. The vendor route is the most common only when registry listings are counted. Among starred GitHub repositories, most publishing servers use your own platform credentials. On the core six platforms (X, LinkedIn, Facebook, Instagram, TikTok, YouTube), 188 of 316 publishing servers sign in through a vendor. Before classification, in our first GitHub search for the 10 platform names with no star floor, the share of repositories pushed to in the last 90 days rose with stars: 51% at 0-4 stars, 54% at 5-9 stars, 57% at 10-49 stars, 63% at 50+ stars. That measure counts pushes to any branch, bots included.
- Platform-built servers. Read on each platform's own developer site and GitHub organization. Pages checked with no MCP server found: LinkedIn's developer documentation on Microsoft Learn and LinkedIn's GitHub organization, Google Cloud's supported MCP products, Google's MCP repository, the Workspace MCP guide, the YouTube Data API docs, docs.bsky.app, atproto.com, Bluesky's GitHub organization, the Threads API index and Meta's MCP hub. Reddit's devvit-mcp is built by Reddit but helps developers write Reddit apps and does not act on accounts, so it is not counted.
- Checks. 1,672 candidates were classified by one set of AI agents, then every label was re-checked by a separate set of AI verification agents against the live repository or docs page (the registry snapshot for servers listed only in the registry), and rows touched by later rulings were checked again for consistency. We never called an MCP server's tools or ran one. During planning, two unauthenticated requests reached TikTok's MCP server address and returned errors, and nothing from them is used.
- Limitations. A census of what is listed, not of what is used: stars and listings say nothing about installs. Servers not in the registry and below 5 stars are missing, so the census undercounts what exists. A label reflects what a server's files say, not a test of the server. Hosted servers with no public code cannot be dated.
Social media MCP server FAQs
01 How many social media MCP servers are there?
02 Does X have an official MCP server?
03 Is there an official MCP server for LinkedIn, Instagram or TikTok?
04 How do social media MCP servers sign in to your account?
05 How many social media MCP servers are still maintained?
Cite this study, and press
Free to quote, cite and reuse these charts and data with a link to this page.
Ramonov, S. (2026). MCP Servers for Social Media Automation: 2026 Statistics. Blotato. https://www.blotato.com/research/social-media-mcp-servers
- Charts: servers by platform (PNG), sign-in routes (PNG), maintenance by builder (PNG), social media share of new MCP repositories by month (PNG).
- Data: all 891 servers (CSV) and full dataset with quotes and source links (JSON). Growth section: monthly series (CSV) and gate table (CSV).
- Press and data questions: seo@blotato.com. Sabrina Ramonov, Blotato's founder, is available for comment.
Disclosure: Blotato sells a hosted social media MCP server, one of the vendor servers counted here, so the sign-in and maintenance findings bear on its own product. That is why every label, except those that read not stated, carries the quote behind it: you can check each one against its source.