connect pipedrive mcp
to your ai agent
give your agents read and write access to your pipedrive crm. connect once with the belt cli, and pipedrive tools sit alongside 250+ built-in tools your agents already use.
what your agent can do with pipedrive
once connected, every pipedrive operation becomes a tool your agents can call, with typed parameters and structured results. that covers the full crm surface:
so an agent can pull every open deal in a pipeline, enrich the contact from another source, update the stage, and log a follow-up activity, all in one run, without you writing a single line of pipedrive integration code.
where a pipedrive agent helps
the value isn't access for its own sake. it's the sales work your team stops doing by hand.
reps spend hours keeping pipedrive tidy. an agent logs activities, updates deal stages, and fills in records as work happens, so the crm stays current without anyone clicking through it.
an agent watches your pipeline on a schedule and surfaces deals that have stalled, drafts the next touch, and logs the follow-up, before they quietly slip past their stage.
new leads get created, enriched from your other connected tools, and routed to the right pipeline automatically, so a rep picks up a qualified deal instead of a raw form fill.
skip the filtered views and saved reports. ask the agent "which q3 deals slipped a stage this week?" and let it read pipedrive directly and answer.
how to connect pipedrive to your agent
three commands with the belt cli. you authenticate once, and the connection is available to your agents, flows, and any mcp client pointed at inference.sh.
list available connectors
see every mcp server inference.sh can connect to. pipedrive is in the list.
belt mcp list # slug category auth status # mcppipedrivecom other api_key — # mcpclickupcom other oauth — # todoist other oauth —
connect pipedrive
run one command and paste your pipedrive api token when prompted. inference.sh stores it securely at the platform level, not in a local config file on every machine.
belt mcp connect mcppipedrivecom # paste your pipedrive api token: •••••••• # ✓ connected. token stored and managed by inference.sh
inspect what's available
see the tools pipedrive exposes. each one comes with a typed schema for its parameters and return values.
belt mcp tools mcppipedrivecom # inference.sh discovers every pipedrive tool # with its typed parameters and return schema
let your agent use them
that is it. your agents now call pipedrive tools exactly like built-in tools. nothing to wire up, no separate credentials to pass around.
why connect pipedrive through inference.sh
the pipedrive mcp server is open. the value inference.sh adds is everything around it: auth, management, and making the tools usable by agents out of the box.
credentials handled for you
authenticate once. for api-key servers like pipedrive your token is stored securely; for oauth servers (slack, notion, todoist) inference.sh runs and refreshes the whole oauth flow automatically. no client ids, redirect uris, or token logic to maintain.
managed centrally, not per machine
connections live at the platform level instead of scattered across every developer's local config. a new teammate authorizes once and gets access. when something changes, you update it in one place.
one unified tool catalog
pipedrive tools sit next to 250+ built-in apps and your own composed flows. an agent can generate an image, read a deal, and post to slack in the same run, all from one catalog with an identical interface.
agents use it natively
from the agent's perspective there is no difference between a built-in tool and a pipedrive tool. both are tools with typed parameters and structured results, so you build the logic, not the integration plumbing.
example prompts
drop these into an agent on inference.sh once pipedrive is connected.
what is the pipedrive mcp server
the model context protocol (mcp) is the standard way ai tools talk to external services. the pipedrive mcp server exposes pipedrive's crm operations as mcp tools, so any mcp-compatible client or agent can discover and call them. inference.sh connects to that server on your behalf and manages the credentials, so you skip the setup and go straight to using the tools.
faq
does inference.sh store my pipedrive api token?
yes, securely and at the platform level. you paste it once during belt mcp connect mcppipedrivecom and it is managed centrally instead of living in a local config file on every machine.
can my agents use pipedrive and other connectors in the same run?
yes. connected mcp servers, the 250+ built-in tools, and your composed flows all appear as one unified catalog, so an agent can call pipedrive, slack, and an image tool in a single execution.
do i need to write any integration code?
no. once connected, pipedrive tools are available to your agents exactly like built-in tools, with typed parameters and structured results.
what if the pipedrive connector used oauth instead of an api key?
then inference.sh would run the full oauth flow for you, store the tokens, and refresh them automatically. that is how oauth servers like slack, notion, and todoist work on the platform.
connect pipedrive in one command
give your agents your crm. credentials managed, tools ready, no plumbing.