The social media MCP server for Claude, ChatGPT and Cursor.

19 MCP tools let any AI agent push ideas, draft posts, read the schedule and pull performance from your workspace.

Request a demo

19 tools

Every tool an agent can call, grouped by what it is allowed to do.

Read

list_brandsLists the brands the token can reach.
get_mining_contextReturns the voice guide, pillars, persona and known source refs.
list_ideasLists ideas already pushed or created in Studio.
get_ideaReads one idea by id.
list_content_profilesLists the content profiles available to write from.
list_postsLists posts by status, platform or date, newest first.
get_postReads one post in full, with its approval audit and permalinks.
list_channelsLists connected accounts and what each can post.
get_scheduleShows what is going out this week and whether it will fire.
get_post_performanceReads one post's metrics for each channel it reached.
get_account_performanceReads per-platform totals and follower counts over a range.
get_top_postsRanks published posts by engagement or views.
get_best_time_to_postNames the brand's best hours once enough posts have data.

Standard write

push_ideasPushes drafted ideas into a brand's queue.
create_postWrites a post into the queue, awaiting approval.
update_postEdits an unpublished post and reschedules its bookings.
update_idea_statusMoves an idea between new, shortlisted and dismissed.

Elevated write

approve_postApproves a queued post, needs the elevated tier.
delete_postDeletes a post that never went live, needs the elevated tier.

Connect in a minute

Add the connector in claude.ai or ChatGPT, one command in Claude Code, or OAuth and a token anywhere else.

Ideas from your git log

The CLI reads your local git log and an optional notes vault, and skips ideas it already pushed.

REST API and webhooks

Call the REST API directly, or listen for the 5 events a post can fire.

Agent permissions and limits

Access tiers are chosen on the consent screen, and every call stays brand-scoped and on the record.

Readlist_brands, get_mining_context, list_ideas, get_idea, list_content_profiles, list_posts, get_post, list_channels, get_schedule, get_post_performance, get_account_performance, get_top_posts, get_best_time_to_postReads the workspace and changes nothing.
Standard writepush_ideas, create_post, update_post, update_idea_statusIdeas and posts land in the queue awaiting approval.
Elevated writeapprove_post, delete_postApproves a queued post, or deletes one that never went live.

Brand-scoped and audited

Questions people ask

What is a social media MCP server?

It is an MCP connector that lets an AI agent read your workspace and write posts through the same API Postlyra's own app uses.

Can Claude post to Instagram or LinkedIn with it?

Claude can draft and queue a post for any connected network including Instagram and LinkedIn. It lands awaiting approval, and approve_post needs the elevated tier before it publishes.

Do I have to share my passwords?

No. Access comes through OAuth on a Postlyra screen, or a bearer token from Settings that you can revoke any time.

Is it free?

Yes. MCP, the REST API and the CLI sit in the Developer group and are available on every plan including Free.

Does it work in Claude Code and Cursor?

Yes. Claude Code connects with one command, and Cursor or any other MCP client connects with the same OAuth or bearer token flow.

Can the agent publish on its own?

create_post writes into the queue awaiting approval. Publishing still needs approve_post on the elevated tier.

Can I use the REST API instead?

Yes. The REST API at api.postlyra.com stays available for channels, posts, slots and media alongside MCP.

Can it read my schedule and analytics too?

Yes. list_posts, get_schedule, get_post_performance, get_account_performance, get_top_posts and get_best_time_to_post read data already stored, refreshed nightly.

Point an agent at Postlyra.

See push_ideas and create_post run from a real prompt, awaiting approval.

Request a demo