MCP Server

Model Context Protocol integration for agent runtimes. Give your AI agents native tool access to search, fetch documents, and monitor changes.

What is MCP?

Open standard for agent tool integration

The Model Context Protocol (MCP) is an open standard for connecting AI models to external tools and data sources. Instead of calling REST APIs directly, agents discover and invoke tools through a standardized interface. The RLVNCE MCP server exposes corpus operations as native agent tools.

Available tools

30+ tools across retrieval, corpus management, crawling, monitoring, and catalog

Core Retrieval

Search, fetch, and monitor

search

Full-text search with filters, attribute filtering, and boosts

view_cache

View cached page content by document ID

list_documents

Paginated document list for a corpus

sample_documents

Fetch random documents for spot-checks

list_changes

Change feed since a timestamp with cursor pagination

list_corpora

List available corpora with status summary

Corpus Setup

Create and configure corpora

create_corpus

Create corpus with sources and policies in one call

get_corpus

Full corpus detail - sources, policies, crawl stats

update_corpus

Update corpus name, description, and policies

update_search_policy

Update field weights, boosts, and filters

add_sources

Add source URLs to an existing corpus

remove_source

Remove a source from a corpus

Crawl Operations

Trigger and monitor indexing

trigger_crawl

Start a crawl, returns job ID

cancel_crawl

Cancel a running crawl job

get_crawl_status

Check crawl progress (pages discovered/crawled/failed)

list_crawls

Crawl history with pagination

list_crawl_results

Per-URL outcomes with status filter

list_crawl_errors

Failed URLs with error details

get_crawl_schedule

Next scheduled crawl and overdue status

Monitoring & Observability

Webhooks, usage, and source health

create_subscription

Set up webhook notifications for change events

list_subscriptions

List active webhooks for a corpus

get_usage

Current metered usage counters

get_limits

Plan limits and remaining quota

list_sources

Per-source status, doc count, and last crawled

get_source_stats

Detailed source stats with failure tracking

list_source_urls

List indexed URLs for a specific source

get_corpus_metrics

Corpus health: coverage, freshness, success rate

Catalog & Subscriptions

Discover and subscribe to published corpora

list_catalog

Browse published corpora with category/search filtering

subscribe_corpus

Subscribe to a published corpus for read-only access

unsubscribe_corpus

Remove a corpus subscription

Quick start

Multiple ways to connect

terminal

# Hosted mode (OAuth - recommended)

claude mcp add rlvnce --transport http https://api.rlvnce.com/mcp

# Local mode (API key, stdio transport)

claude mcp add rlvnce -- npx @rlvnce/mcp-server

# Plugin (tools + workflow skills)

claude plugin install @rlvnce/mcp-server

Hosted (HTTP + OAuth)

OAuth authentication is automatic - you'll be prompted to authorize on first use. No API key needed. Works with Claude catalog.

Local (stdio + API key)

Runs as a local process. Set RLVNCE_API_KEY env var. Works with Claude Desktop, Cursor, Windsurf, and any MCP-compatible IDE.

Integration

Works with any MCP-compatible runtime

Claude

Install via claude mcp add or add the URL to your Claude Desktop or Claude Code configuration. OAuth handles authentication automatically.

Other runtimes

Point any MCP-compatible agent runtime to https://api.rlvnce.com/mcp. The protocol is open - no vendor lock-in.

Give your agents native search

Connect your agent runtime to RLVNCE via MCP.

Get started