Developer Update•Sep 17, 2026

Linxa MCP Gets a Significant Expansion

The latest push adds new methods covering outreach management, prospect search, message sending, and analytics—accessible to any MCP-compatible AI client.

When we launched the Linxa MCP server in February, it gave Claude and other MCP clients read access to your LinkedIn inbox and conversation context. That was a useful starting point for building research and analysis workflows.

The latest push substantially expands what those clients can do. You can now instruct Claude—or any other MCP-compatible agent—to manage outreach sources, search for prospects, send messages, and pull analytics, all through the same protocol.

New methods in this release

Prospect discovery

search_prospects

Search LinkedIn profiles against saved ICP criteria. Returns ranked matches with enrichment data.

get_prospect_profile

Fetch full enrichment data for a specific LinkedIn profile by URL or member ID.

list_sources

List configured outreach sources with their current status and prospect counts.

create_source

Create a new outreach source from a search query or saved search definition.

Outreach actions

queue_invite

Queue a connection invite for a prospect. Respects daily limits and safety windows configured in Smart Invite.

send_message

Send a message to an existing connection. Returns delivery status.

attach_sequence

Attach a named sequence to a source. Messages begin on acceptance per the sequence schedule.

pause_source

Pause or resume an outreach source without deleting its configuration.

Inbox and conversations

list_conversations

List inbox conversations with filter options: unread, by label, by date range, or by growth agent.

get_conversation

Fetch a full conversation thread including all messages and associated prospect data.

update_conversation_label

Set or clear a conversation label: qualified, not-interested, replied, archived, or custom.

get_unanswered

List conversations where the last message was outbound and no reply has arrived.

Analytics

get_invite_stats

Return acceptance rates, pending invites, and daily send counts for a date range.

get_reply_stats

Return reply rates by sequence, message variant, or time window.

get_funnel_summary

Return the full funnel: prospects → invites sent → accepted → replied → qualified.

Signals tracking

list_tracked_authors

List authors in your Signals watchlist, separated by Clients and Competitors.

get_author_posts

Fetch recent posts from a tracked author with engagement counts.

list_saved_searches

List saved Signals keyword searches and their last refresh time.

get_saved_search_results

Return the most recent results for a saved search.

Safety controls still apply

Methods that take action—queue_invite, send_message, attach_sequence—are all subject to the same rate limits, daily caps, and safety windows you configure in the Linxa dashboard. An AI client cannot bypass those limits by calling MCP methods directly. If the daily invite cap is reached, queue_invite returns an error with the remaining capacity and next-available time.

Example: funnel report via Claude

// Claude prompt
"Give me a summary of this week's outreach funnel 
and flag any sequence with reply rate below 10%."

// MCP calls Claude makes
→ linxa.get_funnel_summary({ period: "7d" })
→ linxa.get_reply_stats({ period: "7d", group_by: "sequence" })

// Claude returns a natural-language report
// with the flagged sequences and specific numbers

Upgrading

If you are already running the Linxa MCP server, update to the latest version with npm install -g @linxa/mcp@latest. The new methods are available immediately—no additional configuration required. Your existing API token works.

If you are setting up MCP for the first time, the setup guide walks through everything from token generation to Claude Desktop configuration in under five minutes.

Add Linxa to your AI agent stack

Read the updated method reference and start building custom LinkedIn workflows with Claude, Cursor, or any MCP client.

View MCP Docs Setup Guide