MCP · Open source
Marketing MCP for OpenClaw
OpenClaw skills are MCP servers, so Spreeflo is not an integration to build — it is a skill you enable. Then your agent can run marketing from any chat app it listens on.
Skills are MCP servers·20+ messaging platforms·Open source
{
"mcpServers": {
"spreeflo": {
"url": "https://mcp.spreeflo.com/mcp",
"headers": { "Authorization": "Bearer <YOUR_API_KEY>" }
}
}
}Writes apply to your live workspace immediately — there's no approval queue here. Keep a human in the loop and check OpenClaw's confirmation settings first.
The agent
What is OpenClaw?
OpenClaw is the most widely used open-source agent harness — a personal AI agent that runs continuously, reaches you across more than twenty messaging platforms, and acts proactively on a heartbeat schedule. Its skill system is built on MCP, so adding Spreeflo is the same motion as adding any other skill.
Skills are MCP servers
Enabling a skill means connecting an MCP server. Spreeflo needs no adapter, wrapper, or plugin written for it — the protocol is the integration.
Wherever you already talk
WhatsApp, Telegram, Slack, Discord and more, so “schedule the newsletter” can be a message rather than a login.
Proactive by design
The heartbeat scheduler lets it act without being asked — useful for upkeep, and the reason to be deliberate about what you grant it.
Setup
Connect OpenClaw in about a minute
A Spreeflo API key, and one entry in OpenClaw.
Create an API key
In the Spreeflo app, go to Settings → API Keys and click + Add API Key. Name it OpenClaw so you can revoke this one connection later without touching the others. It goes in the Authorization header — never in the server URL.
Add the Spreeflo MCP server
Skill (MCP server)- 1.Add Spreeflo as an MCP server in your agent configuration — via the visual configurator’s Skills section, or
openclaw config set. - 2.Use the Streamable HTTP form below, with your key in the
Authorizationheader. - 3.Restart the gateway so the new tools are discovered, then confirm the agent can see them.
{
"mcpServers": {
"spreeflo": {
"url": "https://mcp.spreeflo.com/mcp",
"headers": { "Authorization": "Bearer <YOUR_API_KEY>" }
}
}
}Keep the key in an environment variable rather than the config file, and give OpenClaw a key of its own so you can revoke it without touching your other agents.
Check it works
Ask OpenClaw for something small — "list my Spreeflo audiences". A working connection is one where it replies in whichever chat app you asked from, with your real audience names.
Full setup guide and troubleshootingIn practice
What OpenClaw does with it
Two ordinary requests, and the Spreeflo tools behind them.
Draft the product update email and tell me who it would go to.
Drafted the template and previewed the audience it targets:
get_business_profilecreate_email_templateget_segment_preview
Draft only — there is no MCP tool to start the send, so it waits for you in the app.
Each morning, check yesterday’s form responses and tag anyone who asked about pricing.
Runs on the schedule, reads the new responses, and tags the contacts behind them:
get_form_summarylist_tagstag_contacts
A recurring write running unattended — exactly the case to give its own key and review first.
Running it
Living with an agent that acts on its own
OpenClaw is unusual among MCP clients in that it does not wait to be asked. The heartbeat scheduler and the chat surfaces change what granting marketing tools actually means, so these are the parts worth deciding before you enable the skill.
The heartbeat is where writes go unwatched
Anything you put on a schedule runs with nobody reading the output. That is fine for reads and reporting; for writes it means a mistake repeats on every tick rather than once. Start scheduled tasks read-only, watch a few cycles, and promote them to writing once the shape is right.
Chat instructions leave no diff
A request typed into WhatsApp or Telegram has no commit, no review, and no trail on OpenClaw’s side. The record of what actually changed lives in Spreeflo’s audit trail instead, attributed to whoever owns the key — which is a good reason not to share one key across people.
New tools need a gateway restart
Skills are discovered when the gateway starts, so adding the server to your config is not enough on its own. If the agent insists it has no Spreeflo tools, restart before you go looking for a bad key — that is the usual answer.
One key, revocable on its own
Because OpenClaw runs continuously and reaches many surfaces, it is the client most worth giving a dedicated key. Revoking it later stops the agent without disturbing your other integrations, and it keeps the audit trail readable.
A proactive agent will use everything you grant it
OpenClaw is designed to act without prompting, so the tools you enable are not a menu it consults occasionally — they are standing capability. Spreeflo has no MCP tool for the irreversible actions, no matter what the agent asks, but everything short of that (drafting, segmenting, tagging, editing) is fair game on a schedule. Decide what you are comfortable with it doing at 4am, because that is when it will.
Why this pairing
OpenClaw + Spreeflo
- Spreeflo is a skill, not a custom integration to build
- Reachable from the chat apps OpenClaw already listens on
- Heartbeat schedules turn marketing upkeep into background work
- Its own Spreeflo key, revocable without touching other agents
- No MCP tool for deletes, sends or publishes, whatever the agent asks
Coverage
Everything it can touch
The same surfaces you work with in the app, exposed as tools — and a short list that is deliberately off limits.
OpenClaw tends to live in the reporting and upkeep corners of this — performance numbers pulled into a chat thread, tags and segments kept current on a schedule.
Audience
- Audiences and contacts
- Segments, static and dynamic
- Tags
- Custom contact attributes
- CSV import
Sending
- Campaigns and journeys
- Email templates
- Web push templates
- Sender and domain identities
Capture
- Forms and question types
- Inline and popup embeds
- Response summaries
- CSV export of responses
Insight
- Campaign and web push performance
- Tracked events and pages
- Link clicks
- Form analytics
What MCP can’t do
These are irreversible or they reach real people, so they are removed from the surface rather than left to your client's confirmation settings. A person does them in the app.
Supercharge your marketing automation
FAQ
Questions, answered.
No. OpenClaw’s skill system speaks MCP directly, so Spreeflo is enabled as a server rather than installed as code. There is nothing published to a registry, nothing to keep updated, and no adapter maintained in between — when Spreeflo adds tools, your agent sees them after the next gateway restart.
No. The connection is outbound: your gateway dials Spreeflo’s hosted server. You do not need a public address, a tunnel, or an inbound firewall rule, which is why this works the same on a laptop as it does on a VPS.
Yes, and it is one of the better reasons to pair them — weekly reporting, tagging new form responses, keeping a re-engagement segment current. Bear in mind the schedule runs unattended, so favour reads and low-stakes writes, and give the agent a key you can pull if a recurring task misbehaves.
No. Spreeflo runs a hosted MCP server, so OpenClaw calls it directly over Streamable HTTP — nothing to install, nothing to self-host, no SDK to keep up to date. A server URL and a sign-in (or, for a headless agent, an API key in the Authorization header) are the whole integration.
It cannot delete anything or start a send. Spreeflo has no MCP tool for deletes, starting or resuming a campaign or journey, sending or scheduling a broadcast, or publishing a form, rather than leaving them to client settings. Those stay in the Spreeflo app, where a person does them. Journeys you have already switched on keep running, though, so editing a contact, its tags, a segment or an email template, or importing contacts, can set one off, just as the same edit would in the app.
The key resolves to exactly one workspace, and every action runs as the person who owns that key — their membership re-verified on each request. The audit trail names a real person, so agent-made changes are attributable.