Clarify
by Clarify
Query your CRM, import leads, build campaigns and create autonomous GTM agents from your AI assistant. 30 documented tools, 18 of which write. OAuth sign-in with no read/write scope split.
Verified connector
Listed by Anthropic as a partner connector in its Connectors Directory.
Connection checked by Agentman on .
Anthropic states this reflects the level of review a connector received, not a security audit.
Connect Clarify via MCP
https://api.clarify.ai/mcpWorks in any MCP-compatible client. In Agentman Studio it is one click — no config file to edit.
Clarify Tools & Capabilities (26)
Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.
Frequently asked questions
Why does a Clarify write tool return "Tool not found"?
Your client is holding a stale tool list. Clarify renamed several write tools, so `create-record`, `update-record` and `create-list` now fail with MCP error -32602 while read queries keep working. Clarify's own fix is to remove the Clarify app from your client and add it again, because reconnecting alone does not refresh the list.
Can the Clarify connector send email to my prospects?
Not directly. Clarify's MCP tools create campaigns in draft mode, and Clarify documents a connected email account as required to enable a campaign rather than to create one. No tool in either published list enables a campaign. Sending starts when a human flips the Enabled toggle in the app.
Can an agent created through MCP act on its own afterwards?
Yes, and this is the sharpest risk on the connector. Clarify states that agents Rep builds are enabled on creation and start running when their trigger next fires. An agent can send email automatically when configured to, and Clarify names the absence of undo for agent actions as its most significant limitation today.
Does the Clarify connector respect my CRM permissions?
Partly, and the granularity is coarse. Clarify scopes agents to their creator's permissions, so an agent cannot read records its owner cannot. But Clarify offers only two roles, Admin and Member, and states more granular permissions are planned. Members reach every CRM record, so role is a weak boundary.
Why does my assistant answer without checking Clarify?
Your AI client decides when to call a connector, not Clarify. Clarify recommends three fixes: confirm Clarify is enabled for that specific conversation rather than only added to your account, name Clarify in your first message, and add a line to your project instructions telling the assistant to use Clarify for CRM questions.
How do I use the Clarify MCP server on a schedule?
Authenticate with a Personal API key instead of OAuth. Clarify states the interactive login needs periodic re-authorising, which interrupts scheduled work. Send the key as an `api-key` authorization header alongside an `X-Clarify-Workspace` header. Only Personal keys work; Clarify warns a workspace key lists every tool then fails on every call.
Can I approve Clarify writes one at a time instead of all at once?
Only in your AI client, not at Clarify's consent screen. Clarify's OAuth server advertises four scopes and all four are identity scopes, so there is no read-only grant to choose. Clarify states every tool is available to all workspaces with nothing to enable on its side, which puts the whole approval burden on your client.
Sources
- Clarify MCP documentation, for the 30-tool list and its read/write grouping, the per-client setup steps, the Personal API key route, and all seven troubleshooting rows used in the FAQ (retrieved 2026-08-18; also available as clean Markdown via the
.mdsuffix). Page byline states it was updated the day before retrieval. · retrieved 2026-08-18 - Clarify Agents, for enabled-on-creation behaviour, the trigger table, human-in-the-loop approval, creator-scoped permissions, the no-undo and no-cancel limitations, the 100-runs-per-hour cap, the five-hop recursion break, the per-integration can/cannot lists, credit costs and the 30-step historical processing cap (retrieved 2026-08-18) · retrieved 2026-08-18
- Clarify Campaigns, for draft-versus-enabled semantics, the plan limits table, the 1-credit-per-send charge, auto-enrollment, the removal-does-not-stop-sending note, variable resolution at send time, workspace-level unsubscribes, and the immutable-once-enabled and creator-only-edit rules (retrieved 2026-08-18) · retrieved 2026-08-18
- Running outbound with Clarify, for the 250-emails-per-day-per-account sending cap, the connected-email prerequisite and the agent-drafts-versus-sends distinction (retrieved 2026-08-18) · retrieved 2026-08-18
- Clarify Roles & Permissions, for the two-role model, the Admin-only API key and custom MCP controls, the all-users-were-Admin default in pre-existing workspaces, and the stated plan for more granular permissions (retrieved 2026-08-18) · retrieved 2026-08-18
- Clarify API — AI Agents, for the
api-keyauthentication scheme, the workspace header, the Personal-keys-only rule, the REST base URL and the OpenAPI spec (retrieved 2026-08-18) · retrieved 2026-08-18 - Clarify API — Rate limits, for the 3,000 requests per minute per workspace per endpoint figure, the rolling 60-second window and the rate-limit response headers (retrieved 2026-08-18) · retrieved 2026-08-18
- Live auth posture check against
https://api.clarify.ai/mcp— an anonymousinitializereturned HTTP 401 with awww-authenticateheader carryingresource_metadataand the four scopes;/.well-known/oauth-protected-resource/mcpand the host-root form both returned HTTP 200 with identical bodies while the path-append form returned 404;/.well-known/oauth-authorization-serverreturned HTTP 200 at the host root and 404 at the path-insert form;auth1.clarify.aiserves the same document plus an OpenID configuration (2026-08-18) · retrieved 2026-08-18 - Tool-list diff — Anthropic's 26 snapshot names and Clarify's 30 documented names were set-differenced programmatically on 2026-08-18, giving 18 shared, 8 directory-only and 12 documentation-only entries · retrieved 2026-08-18
docs.clarify.ai/robots.txtdisallows only/not-authorized,/not-foundand a template path, setsCrawl-delay: 1, namesdocs.clarify.ai/sitemap.xml, and carries noContent-Signalheader on any axis.developer.clarify.ai/robots.txtisAllow: /.docs.clarify.ai/llms.txtreturns HTTP 200;www.clarify.ai/llms.txtreturns 404 (all checked 2026-08-18) · retrieved 2026-08-18- Anthropic Connectors Directory entry, for the 26 tool names, the partner tier, the empty
prompt_namesarray, the streamable-http transport and the "Read and write" permissions field (snapshot 2026-08-16) · retrieved 2026-08-16 - Clarify support — <mailto:support@clarify.ai> · Privacy
Use in Agentman
Connect once and your agents call these tools on their own — on a schedule, in a workflow, with nobody at the keyboard.
Open in Agentman StudioServer Info
- Category
- Productivity
- Developer
- Clarify
- Tools
- 26
- Domain
- api.clarify.ai
Using Claude Desktop or another MCP client? Setup docs — the connection URL above works anywhere.