1Social

1Social MCP is aremote, OAuth-securedserver. There is no package to install and no API key to paste. Add this to your client's MCP config:

{ "mcpServers": { "1social": { "url": "https://mcp.1social.dev/mcp" } } }

- 📤 Publish and schedule— one message goes to every network you pick, now or up to 30 days ahead.
- 🔍 Check before it goes— per-network character counts, whether an image is required, whether a multi-part post will thread natively or be joined, and what it costs against your quota.
- 🧾 Read the receipt— what reached each network, what failed and why, and what genuinely could not be determined.
- ♻️ Fix what broke— retry one network without touching the ones that already went out.
- 🖼 Attach media— post an image or video from the chat. Instagram and TikTok refuse posts without one.
- ✏️ Edit, cancel, delete— change a scheduled post, unschedule it, or delete it from the networks that allow deletion.

Ten tools, deliberately few. Publish-now and schedule are the same tool (scheduledAt: nullmeans now), and reading one post islist_postswith apostIdrather than a second tool — a smaller set is one a model picks from accurately.

The live set is what the server says it is — runtools/listagainsthttps://mcp.1social.dev/mcpfor the exact, current shapes.

A post to four networks can land on three. Every tool result is per network: the failure is named in words you can act on, and a delivery whose outcome genuinely could not be determined comes back asunknownrather than as success.

unknownis not a dead end. You look at the network, tell the assistant what you see, andconfirm_deliverysettles it — confirming it is live commits the quota, confirming it is not marks the delivery failed soretry_deliverycan pick it up. The assistant is not allowed to guess this on your behalf.

delete_postalways removes the post from 1Social. Removalfrom the networkdepends on the network:

On a multi-part thread only the first post is removed; the rest stay live as orphans. The tool reports what actually came down, per network, and the assistant is instructed to tell you that rather than confirm a clean delete.

A new account has a small free monthly allowance — enough to publish, watch the receipt come back, retry something, and schedule a post — so the tools work before you pay for anything. A failed delivery does not count against the quota when the failure was ours or the network's.

You never have to leave the chat to pay.A tool call that runs out of quota comes back with a checkout link for your account.

- Claude(web & desktop) — Settings → Connectors → Add custom connector
- ChatGPT— Settings → Apps & Connectors → Advanced → Developer mode → Add custom connector
- Cursor— one-click deeplink, or~/.cursor/mcp.json
- VS Code— one-click deeplink, or.vscode/mcp.json
- Claude Code (CLI)—claude mcp add --transport http 1social https://mcp.1social.dev/mcp
- Gemini CLI—gemini mcp add --transport http 1social https://mcp.1social.dev/mcp
- Codex CLI—codex mcp add 1social --url https://mcp.1social.dev/mcp
- Generic HTTP MCP client— point it at the URL and let it complete OAuth discovery

This repository is also a Cursor plugin. It contributes one MCP server (1social) and no rules, skills, agents, or hooks.

.cursor-plugin/plugin.json ← plugin manifest mcp.json ← the MCP server it contributes assets/logo.svg ← plugin logo

There is nothing to configure. The server advertises OAuth 2.1 metadata, Cursor registers itself dynamically, and the consent screen opens on first use.

"Draft a post about our new pricing page for LinkedIn, X and Bluesky, check it fits, then publish."

→preview_postreports the per-network counts and cost;publish_postreturns a receipt per network.

"Post this screenshot to Instagram and Threads on Friday at 9am."

→ The attached file goes inmedia; Instagram would have refused the post without it.

"Did yesterday's post make it everywhere?"

→list_postsreturns the receipt — including any delivery still markedunknown.

"LinkedIn failed. Retry just LinkedIn."

→retry_deliverytouches one delivery and leaves the successful networks alone.

OAuth 2.1 with PKCE, per the MCP authorization spec. The authorization server ishttps://1social.dev; it supports Dynamic Client Registration, so compatible clients set themselves up and you only see the consent screen. Scopes granted areopenid,profile,emailandoffline_access.

The assistant never sees your social credentials.Your network tokens stay in 1Social; the assistant holds only a token for your 1Social account, and you can revoke it at any time from the web app.

This repo is the source for two listings:

Thetoolsarray inlhm.plugin.jsonisgeneratedfrom theTOOLStable in the server's own contract (packages/contract/src/mcp/tools.tsin the private repo), the same table that answerstools/list. It is a snapshot: when the tool surface changes, regenerate it rather than hand-editing, or the listing will describe a server that no longer exists.

Note that the two registries carry different version numbers on purpose —server.jsonis at0.1.0, while LobeHub assigned1.0.0at import time and a version cannot be renamed.

The 1Password MCP server creates a bridge that allows MCP clients such as Codex and Kiro to manage your 1Password Environments with secure authorization prompts.

This is the 1st, easiest, and cheapest PPT, slides, presentation AI generation MCP Server in the world.

Persistent memory for any AI assistant. Zero token cost until recall. Stores memories in local SQLite, ranks by 6-factor scoring, returns results 79% smaller than JSON. Works with Claude, ChatGPT, Grok, Cursor, Windsurf, and any MCP client.

A MCP server that enables AI assistants to interact with Anki, the spaced repetition flashcard application.

Enables LLM clients to interact with macOS applications through AppleScript. Built using the @beyondbetter/bb-mcp-server library, this server provides safe, controlled execution of predefined scripts with optional support for arbitrary script execution.

An MCP server for WordPress plugin audits

Turn your AI assistant into a digital marketing hub that creates, organizes, and analyzes links and QR Codes on demand.

Connect AI clients to Cal.com scheduling through the Model Context Protocol using the hosted server at mcp.cal.com or a local instance.