Developer docs

MCP & Developer ToolsBeta

Bring your own AI. The CaseRescue MCP server lets your firm connect Claude (claude.ai, Claude Desktop, or the API) directly to your workspace's data through the open Model Context Protocol. It is strictly read-only: your AI can ask about your calls and clients, and can never change anything.

What Beta means: the MCP connector works today and is safe to use, but we are still expanding it. Tool names, limits, and capabilities may change as we improve it, and we may occasionally need short maintenance windows. Your data and its read-only protections are not affected. Feedback is welcome and shapes what we build next.

Connect in four steps

Generate a connection token

In the CaseRescue app, go to Settings → Integrations → AI Connector (MCP) and click Generate token. Copy it once; only a hashed version is stored.

Copy your connector URL

The same card shows your workspace's connector URL. Every request is scoped to your workspace; the token can never see another firm's data.

Add it to your AI tool

In claude.ai or Claude Desktop, add a custom connector (remote MCP server) with that URL and your token as the bearer credential. API users pass the same header.

Ask questions

Your AI can now query calls, intakes, complaints, appointments, agent performance and transcripts, and answer questions grounded in your firm's real data.

Available tools

A fixed allowlist of 16 read-only tools. Nothing outside this set is callable.

query_callsSearch and filter your firm's analyzed calls

query_intakesQuery intake records and outcomes

query_complaintsList flagged client complaints

query_positive_feedbackList gratitude and review candidates

query_appointmentsQuery booked consultations

get_agent_performancePer-agent grades and trends

get_contact_journeyA client's full journey and satisfaction

get_client_360Everything known about one contact

get_client_risk_reportRefund and churn risk across clients

get_workspace_analyticsWorkspace-level metrics and trends

get_case_manager_analyticsCase-manager task and promise analytics

query_case_manager_tasksOpen follow-ups and promises made on calls

search_call_transcriptsFull-text search across transcripts

get_call_detailsOne call's full analysis

get_call_insightsAggregated insights across calls

aggregate_dataCustom aggregations over your data

Security model

Read-only by design: no tool can create, modify, or delete anything.
Tenant isolation: every token is scoped to one workspace and can never see another firm's data.
Tokens are stored hashed, shown once at creation, and revocable at any time in Settings.
Rate limited to 60 requests per hour per workspace.
Every tool call is logged with the token that made it.

Want API access beyond MCP?

Tell us what you're building. Webhooks and deeper integrations are available for firms on the platform.

Talk to us ➜