Alpic MCP server icon

Alpic

by Alpic

HIPAA CompliantSOC2 ReadyISO 27001 Ready
Productivity5 tools

Browse the MCP servers you host on Alpic, read deployment logs and pull usage analytics from an AI agent. Anthropic lists 5 tools, Alpic documents 4, and every one of them reads. OAuth sign-in with no application scopes.

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 Alpic via MCP

https://mcp.alpic.ai/mcp

Works in any MCP-compatible client. In Agentman Studio it is one click — no config file to edit.

Alpic Tools & Capabilities (5)

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Alpic's own tool list is one name shorter than its directory entry. Alpic documents four tools; Anthropic lists five. show-analytics appears only in the directory. We could not handshake to settle it, so the count is bounded rather than known.
  • No application scopes exist. The authorization server advertises openid and nothing else, and the RFC 9728 resource descriptor declares no scopes_supported at all. Nothing at the OAuth grant separates reading from writing — the read-only property comes entirely from which tools are exposed.
  • The same credential type drives the full REST API. Alpic documents its bearer scheme as accepting an API key or an OAuth access token, and that API includes deleting a project and all its environments. No MCP tool reaches those operations today.
  • We could not read tool schemas or safety annotations. The endpoint returned 401 to an anonymous request, so no parameter detail, readOnlyHint or destructiveHint is published here. The read classification above is ours, from the names and Alpic's descriptions of them.
  • Analytics retention caps how far back you can look. Alpic documents seven days on Free, thirty on Pro and unlimited on Enterprise. Ranges beyond your plan appear locked.
  • Free-plan request limits are counted but not yet enforced. Alpic documents a request-limit-exceeded reason code returned as HTTP 402 with JSON-RPC error -32002, and notes on the same page that it is not returned today. Because initialize goes through the same check, an over-limit team will fail at connection time once enforcement starts, and most clients render that as a generic outage.
  • Tool calls time out at 30 seconds on Alpic-hosted servers generally, per its troubleshooting guide. That is a platform limit affecting servers you host, and we found no statement either way about whether Alpic's own MCP server is exempt.
  • User counts undercount non-ChatGPT traffic. Alpic states only ChatGPT supplies a client-provided subject, so unauthenticated traffic from Claude or Cursor is never counted as users. A low Users number against high Sessions usually means unauthenticated traffic, not low usage.
  • Whether show-analytics writes anything is unresolved. We classified it as a read from its name and its siblings, with no schema and no vendor description to check against. Treat that as an inference, not an observation.

Frequently asked questions

No. All five tools listed read. None of them deploys an environment, edits a variable or deletes a project, even though Alpic's REST API offers all three. The tools are a deliberately narrow read slice of that API, so the connector inspects what you host without being able to change it.

Anthropic's directory lists five and Alpic's documentation table lists four. The extra name in the directory is show-analytics, which Alpic's page never mentions. Since the endpoint requires authentication we could not run a handshake to settle which list is current, so treat five as the ceiling and four as the documented floor.

Only openid. Alpic's authorization server advertised exactly one supported scope on 2026-08-22, and its own CLI asks for openid, email and profile. All three are OpenID identity claims. No scope distinguishes reading a project from changing one, so consent is all-or-nothing over whatever the token can reach.

Not by default, and only if you opted in. Alpic records metadata only unless you enable session replay plus payload collection, which are off by default and paid-plan only. Where you have enabled them, analytics a tool reads can include request and response bodies your own users generated.

Yes, over streamable HTTP at https://mcp.alpic.ai/mcp. The server returns a WWW-Authenticate header pointing at its RFC 9728 descriptor and advertises a registration endpoint, so a client that supports dynamic client registration can complete the OAuth flow without a pre-registered client ID.

Whatever your build printed. Alpic's API returns build log lines as raw strings with a timestamp and no redaction layer described in its documentation. A build that echoes a token or a connection string puts that text into the agent's context, so treat log output as untrusted and unredacted.

Yes, on any plan including Free. Alpic's Free plan includes 10,000 requests per month across the whole team, and every MCP request counts, including initialize and tools/list. Analytics retention is seven days on Free and thirty on Pro, which caps how far back a tool can read.

Because Alpic classifies auth at deploy time, not continuously. Alpic sends an unauthenticated initialize request during deployment and marks a server public unless it answers 401 with a correctly formatted WWW-Authenticate header. Fixing the header alone is not enough — the classification only re-runs on your next deploy.

Sources

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 Studio

Server Info

Category
Productivity
Developer
Alpic
Tools
5
Domain
mcp.alpic.ai

Using Claude Desktop or another MCP client? Setup docs — the connection URL above works anywhere.