MCPify Slack
Pricing
from $0.05 / actor start
MCPify Slack
Turn your Slack into an MCP client for your Apify tools and any MCP server. DM or @mention the bot — it runs your own tools and replies in-thread.
Pricing
from $0.05 / actor start
Rating
0.0
(0)
Developer
inovaflow
Maintained by CommunityActor stats
0
Bookmarked
1
Total users
1
Monthly active users
4 hours ago
Last modified
Categories
Share
Talk to your own Apify tools — and any MCP server — right inside Slack. DM the bot or @mention it in plain English, and it runs your tools and answers in the thread.
If your team already lives in Slack, you've probably wished you could just ask for things — "pull the latest leads," "scrape this page," "check that listing" — without leaving the chat to go run a tool somewhere else. MCPify does exactly that. It connects your MCP connectors (the Apify MCP server and any others you've saved) to Slack, so the bot can use your real tools to answer.
Here's the part that's usually painful: to do this yourself, every team would have to build and maintain its own Slack app and wire up the gateway and backend that sit between Slack and your tools. We did that once. MCPify is one shared Slack app — the gateway and backend are already built and hosted by us — so your workspace just installs it in a single click, and the rest is handled. You don't touch Slack app config, OAuth, or event plumbing at all.
Who it's for
- Teams on Apify who want to trigger their Actors and MCP tools by chatting, not clicking.
- Anyone who lives in Slack and would rather DM a bot than open another tab.
- Privacy-minded teams who want the AI to run on their account with their keys — never a shared, multi-tenant toolbox.
How it works
There are three pieces, and only the middle one is shared:
- The MCPify Slack app — one app you add to your workspace. It receives your Slack messages and posts the replies.
- Our control plane (hosted by us at
mcpify.inovaflow.app) — the small service in the middle. It handles the one-click install, remembers which workspace is paired to which Actor, and routes messages. - This Actor — your own copy, running on your Apify account. It does the real work: it connects your MCP tools and runs the Claude tool-use loop to produce the reply.
What actually flows between them: when you @mention the bot in a channel or DM it, that Slack message goes to the MCPify app → to our control plane → forwarded to your Actor. Your Actor runs your tools, produces the reply text, and sends it back the same way; we post it into Slack. That's the whole loop.
Security & privacy
- We don't store your messages. Your Slack messages, the surrounding thread, and the bot's replies are never written to our database or disk. For each message we read the thread live from Slack, pass it through in memory to your Actor, post the reply, and discard it.
- Your tools and AI key never leave your account. Your MCP connectors and your Anthropic key live only in this Actor, on your Apify account — they're never shared with us or anyone else.
- What we keep is the minimum to run the app, encrypted. Just two records on our control plane: your install (team ID/name + Slack bot token) and your pairing (your Actor's address + a scoped Apify token to wake it). Both are encrypted at rest (AES-256-GCM); the Slack token is used only to read those messages and post replies.
How to set up
📹 Prefer to watch? Step-by-step setup video
You'll need a free Apify account and the MCPify app in your Slack. Four steps.
- Add the MCPify app to your Slack workspace. Click Add to Slack at https://mcpify.inovaflow.app/slack/install — that works for any workspace, with no Slack Marketplace listing needed. It runs the standard Slack OAuth flow and adds the bot to your workspace.
- Once you authorize, your setup code (looks like
MCP-XXXX-XXXX-XXXX) is shown right there — keep it private, since anyone who has it before your instance registers could pair their own. If you missed it, run/mcpify pairin Slack to see the same code again.
- Once you authorize, your setup code (looks like
- Open this Actor on Apify and create a Task. (Create a Task — don't just hit "Start" on the bare Actor; only a Task's saved input reaches the Standby run.) In the Task input: paste the setup code, add an Apify API token, and pick your MCP connector(s) (and a model; optionally your own Anthropic key).
- Turn on Standby for the Task and Start it — once. That one Start registers the Task with MCPify. After that you don't keep a run alive: even if the run stops, Standby keeps serving as long as the Task stays saved — the platform wakes it on demand when a message arrives.
- @mention the bot in a channel, or DM it. It uses your tools and replies in the thread.
Input
| Field | What it is |
|---|---|
| Slack setup code | The MCP-XXXX-XXXX-XXXX code from /mcpify pair. Links this Actor to your workspace — first run only. Keep it private. |
| Apify token | Lets MCPify wake this Actor when you message the bot (see the FAQ — this is not your connectors' token). A standard Apify API token works; if you scope it down, make sure it can still start this Actor. Stored encrypted. |
| MCP connectors | Your saved MCP connectors (Console → Settings → API & Integrations → MCP Connectors), e.g. the Apify one at https://mcp.apify.com. Pick several and MCPify merges their tools. At least one required. |
| Model | Haiku (fast & cheap, default) or Sonnet (smartest for complex, multi-step tasks). |
| Anthropic API key (BYOK) (optional) | Leave blank to use MCPify's shared key (metered by token — see Costs; paid Apify plans only). Or bring your own key, and the AI is free. |
Conversation memory
The bot remembers the conversation. On each message, MCPify reads the surrounding Slack thread (or recent DM history) and gives it to this Actor as context, so you can say "now book the second one" or "do that again for next week" and it knows what you mean.
- Nothing is stored. That thread is read live from Slack each time and passed through in memory — MCPify keeps no copy of your messages or the replies (see Security & privacy above).
- Security: only your own messages and the bot's own replies are used as memory. Other people's messages — and anything posted by other apps in the same thread — are dropped, so nobody can slip instructions into a bot that's holding your connectors.
- DMs work out of the box; for channel memory the bot just needs to be in the channel (it is — that's how you @mention it).
Slack commands
/mcpify pair— get your private setup code + the activation steps./mcpify status— check whether your instance is connected./mcpify unpair— disconnect this workspace (installer only; also rotates the setup code, then points you back to/mcpify pairto reconnect)./mcpify help— list the commands.
Costs
MCPify runs entirely on YOUR Apify account, so the bill is yours and fully transparent. You pay only for what you use:
| Charge | When | Price |
|---|---|---|
| Standby compute | Keeps the bot reachable; scales to zero when idle | Apify's usual compute rate |
| Tool runs | Any Apify Actor your tools trigger | bills to your account as usual |
| Actor start | Each time the bot wakes from idle — about once per run, not per message | $0.05 |
| AI tokens (shared key) | Per reply, by model — Haiku $1 / $5 per 1M tokens (in / out), Sonnet $3 / $15 | metered per use |
The token charges are Anthropic's standard per-million-token rates, passed through at essentially zero margin — Haiku is cheap (a typical question is a cent or two; Sonnet costs more). Bring your own Anthropic key (BYOK) and the AI tokens drop to zero — you pay Anthropic directly; only the start fee and compute/tool runs remain.
Tip: set a Maximum charge per run to cap spend — when it's hit, the bot asks you to raise it or switch to your own key.
FAQ
Why does it need an Apify token? Only so MCPify can wake and reach this Actor when you message the bot (Apify's Standby gate needs an authorized caller). It is not used for your MCP connectors — those run on the token Apify injects into every run for free. Think of the pasted token as the "wake my Actor" key. It's stored encrypted.
Why create a Task instead of just running the Actor? A bare Actor's Standby runs receive no input, so it couldn't know your setup code or connectors. A Task saves your input and hands it to the Standby run — that's the supported way to run a configured Standby Actor.
Do I have to keep a run going? No. You Start the Task once to register it; after that, Standby serves on demand as long as the Task stays saved. The platform wakes it when a message arrives and lets it sleep when idle.
Is the first reply slow? If the Actor has scaled to zero, the first message wakes it and takes a few seconds (~6s) while it boots; after that, replies are instant. MCPify waits for the boot to finish rather than failing the first ping.
Notes
- This Actor runs in Standby mode (default). It exposes no public API of its own — it only accepts events signed by the MCPify app.
- It never sees your Slack token. It returns reply text; the MCPify app posts to Slack.
- It runs on your Apify account with your credentials — your data stays yours.