Hosted MCP server · 36 tools

Run your social media
from your AI agent.

Triage drafts, draft a week from live trends, design covers, and schedule to the best slot — right from Claude, Cursor, or whatever you're already in all day. Everything still waits for your yes.

Join the waitlist →See the full API

# connect from Claude Code — one line, no install
claude mcp add --transport http krevaya https://api.krevaya.com/mcp \
    --header "Authorization: Bearer krv_..."

Connected in three steps

Mint a key

Settings → API keys. Scoped, shown once, revocable — a read-only key gives your agent eyes but no hands.

Add the server

One command (above) or one config entry in any MCP client. Remote and stateless — nothing to install or keep running.

Ask in plain language

“What's waiting for review?” “Draft next week from what's trending.” Drafts land in your queue — you still tap yes.

What your agent can do

Triage the queue

Read what's waiting, approve the good ones, reject the misses, tighten a caption before it ships.

Draft a whole week

One prompt fans out into a week of scored drafts in your voice — from live trends or a URL you hand it.

Design the visuals

Apply any of the 52 studio templates, draw custom graphics, and attach them to posts.

Schedule smart

Ask for the next best slot — the same best-time model the dashboard uses picks the moment.

Read the results

Analytics, the Coach's weekly audit, and trend intelligence come back as data your agent reasons over.

Stay in bounds

Tools are filtered per key scope; drafts wait for your approval; admin tools never appear for normal keys.

Tested with

Claude CodeClaude DesktopCursorCodex CLICopilot · VS CodeGemini CLIWindsurfClineRoo CodeZedOpenAI Agents SDK

All 36 tools

Every tool your agent sees, exactly as the server advertises it — filtered per key, annotated so clients can auto-approve reads.

Read & triage · 7

whoamilist_postsget_postget_queue_summaryapprove_postreject_postupdate_post

Create · 7

create_postdraft_weekrepurposegenerate_video_scriptrender_faceless_videovideo_statusjob_status

Design · 10

design_postlist_design_templatescreate_drawingupdate_drawingget_drawinglist_drawingsexport_drawingsave_drawing_to_libraryattach_drawing_to_postdelete_drawing

Schedule & publish · 5

schedule_postunschedule_postget_best_timeslist_accountslist_brands

Media · 3

upload_assetlist_assetsdelete_asset

Measure · 4

get_analyticsget_auditlist_trendsget_usage

Agent questions

The Model Context Protocol is the open standard that lets AI assistants call real tools. Krevaya hosts a remote MCP server, so your agent gets your whole content pipeline as tools — no install, no local process, one line of config.

No. Mint an API key in Settings → API keys, paste one connect command (or one config entry) into your client, and start asking in plain language: 'triage my drafts', 'draft next week from what's trending', 'schedule the reel for the best slot'.

Only within the rules you already set. Drafts your agent creates land in the same review queue as everything else, and a read-only key can't publish at all — tools are filtered per key, so your agent only ever sees what its key allows.

Any MCP client that speaks Streamable HTTP with auth headers — Claude Code, Claude Desktop, Cursor, Codex CLI, Copilot in VS Code, Gemini CLI, Windsurf, Cline, Roo Code, Zed, and the OpenAI Agents SDK are all tested.

MCP access is included from the Plus plan up, using the same usage allowances as the rest of your account.

Your agent does the work. You keep the yes.

MCP access is included from the Plus plan up.

Join the waitlist