Open-source AI social media scheduler

Let AI agents write your posts. Approve every one before it ships.

SocialHandoff sits between your AI agents and your social accounts. Claude, ChatGPT, Cursor, or n8n drafts and schedules; you approve from your desk or your phone; then it publishes to X, LinkedIn, Bluesky, Mastodon, Threads, Instagram, Facebook, and TikTok.

14 days free. No card required.

  • Works with any MCP client
  • Approve from your phone
  • Guardrails on every draft
  • Tamper-evident audit log
  • Open source, self-hostable
Open source under AGPL-3.0 Read every line that touches your accounts. Self-host it free, with no channel or agent limits.
Hash-chained audit log Every draft, approval, and publish is appended to a chain you can verify with one API call.
Tokens encrypted at rest Social account tokens use AES-256-GCM. Agent keys are stored only as SHA-256 hashes.
No agent approves itself Approval needs an owner key. An agent's own key can draft and schedule, never approve its own post.

Nothing publishes until a person says yes

Agents draft into an approval inbox. You decide what goes live, what gets edited, and what gets thrown out.

Approve from your desk or your phone

Use the desktop inbox, or swipe through drafts on your phone. The inbox installs as an app.

Edit inline or send it back

Fix a word yourself, reject the draft, or request edits so the agent tries again with your note.

A 60-second undo on publish-now

Changed your mind? Publish-now posts wait 60 seconds before they go out. Dry-run preview shows exactly what each network will get.

Connect the agents you already use

One MCP server and one REST API. Your agent plans the week, drafts posts, and reads results without a custom integration.

Any MCP client, in minutes

Run it over stdio with npx socialhandoff-mcp or over streamable HTTP. Claude, ChatGPT, Cursor, VS Code, and Gemini CLI all connect.

One key per agent, scoped tight

Each agent gets its own key, limited to the networks and permissions you pick: draft, schedule, or analytics.

Releases become drafts

A GitHub release or signed webhook turns into a draft for approval, with a citation back to the source.

Guardrails catch mistakes before you see them

Rules run on every draft, so the inbox only fills with posts worth your time.

Rules that match each network

Banned topics, per-network length, missing alt text, daily limits per account, and blocked link shorteners.

Trust is earned, and lost

New agents are draft-only. After clean approvals you can promote one to auto-publish on specific networks. A guardrail failure demotes it at once.

One kill switch for everything

Pause every agent and hold the whole queue with one action. Resume releases the held posts.

Prove who did what, and when

Every action is recorded in an append-only log you can check yourself, not a history you have to trust.

Hash-chained and verifiable

Each entry is chained to the one before it. Call the verify endpoint to confirm nothing was changed or removed.

Citations for every draft

See where a post came from, such as the GitHub release an agent announced.

Policy lives in your repo

Trust rules sit in .socialhandoff/policy.json, so changes to what agents may do go through code review.

Works with your agents and your automations

Connect over MCP, the REST API, the TypeScript SDK, or the CLI. Then publish to X, LinkedIn, Bluesky, Mastodon, Threads, Instagram, Facebook, and TikTok.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Cline
  • Gemini CLI
  • n8n
  • Zapier
  • Make
  • LangChain
  • CrewAI
  • OpenAI Agents SDK
  • GitHub Actions

Integrations for AI agents and automations

60 seconds

to undo any publish-now post

Approved something too fast? Publish-now posts wait 60 seconds before they reach the network, so one tap takes it back.

How approvals work

Self-host free, or let us run it

The open-source version is free with no limits. Hosted plans start at $19/month with a 14-day trial and no card.

Self-host

The full open-source core on your own server.

$0
$0

  • Unlimited channels and agents
  • Approval inbox, guardrails, and kill switch
  • MCP server, REST API, SDK, and CLI
  • Hash-chained audit log
  • Docker Compose or all-in-one image
  • Community support on GitHub
Self-host free

Starter

For a founder and a couple of agents.

$19 /month
$15.83 /month

billed annually

  • 5 channels
  • 2 agents
  • Approval inbox on desktop and phone
  • Guardrails, trust ladder, and kill switch
  • Brand memory with your own model key
  • Hosted, updated, and backed up for you
Start free trial
Popular

Pro

For teams running agents every day.

$49 /month
$40.83 /month

billed annually

  • 15 channels
  • 5 agents
  • Multi-reviewer approvals
  • Everything in Starter
  • More room for channels and agents
  • Built for daily agent workflows
Start free trial

Agency

For agencies managing many clients.

$149 /month
$124.17 /month

billed annually

  • 60 channels
  • 25 agents
  • Client workspaces
  • Multi-reviewer approvals
  • Everything in Pro
Start free trial

14-day free trial, no card required. Cancel anytime. If a paid charge was a mistake, ask within 14 days for a refund under our refund policy at /refunds/.

SocialHandoff pricing: simple plans for teams and their agents

Questions before you connect an agent

Can an AI agent post to my accounts without me?
Not by default. Every new agent is draft-only, and an agent can never approve its own draft. Auto-publish is off until you promote a specific agent on specific networks, and any guardrail failure demotes it again.
What if I approve a post by mistake?
Publish-now posts have a 60-second undo window. You can also hit the kill switch to pause every agent and hold everything in the queue until you resume.
Which AI agents work with it?
Any MCP client, including Claude Desktop, Claude Code, ChatGPT, Cursor, VS Code, Windsurf, Cline, and Gemini CLI. Automation tools like n8n, Zapier, and Make connect over the REST API.
Which social networks are supported?
X, LinkedIn, Bluesky, Mastodon, Threads, Instagram (professional accounts), Facebook Pages, and TikTok. YouTube, Pinterest, and Google Business Profile are not supported yet.
How are my social account tokens protected?
Tokens are encrypted at rest with AES-256-GCM, and agent API keys are stored only as SHA-256 hashes. Every action is written to a hash-chained audit log you can verify.
Is it really open source?
Yes. The core is AGPL-3.0, including the scheduler, approvals, guardrails, and network connectors. You can self-host it free with no channel or agent limits.
Will the posts sound like my brand?
Brand memory stores your voice rules, products, and banned topics, and agents read it before drafting. You can rewrite drafts in your voice using your own model key.
How much does the hosted version cost?
Plans start at $19/month after a 14-day free trial with no card. Annual billing gives you 2 months free, and self-hosting is free.

SocialHandoff FAQ: AI social media answers

Put your agents to work. Keep the last word.

Start a 14-day hosted trial with no card, or self-host the open-source version free today.

14 days free. No card required.

The hosted version is opening soon

Leave your email and we will send one message when hosted accounts open, with your 14-day trial ready. Want it today? Self-host it free from the repo.

Start free trial