Free AI-powered LinkedIn inbox management: search conversations, filter by sentiment and intent, track leads, and take action — all without sharing your LinkedIn password. No paid plan required.
Try these with your AI agent:
Who messaged me on LinkedIn this week?
Show my hot conversations with positive sentiment
Find all messages about hiring
List investors who reached out to me
What are my next actions on LinkedIn?
Show the full thread with [person name]
Add a note to John: "Follow up after demo on Friday"
Mark my conversation with Sarah as read
List conversations labeled "Need Follow Up" with intent direction to_me
Search LinkedIn messages for "partnership proposal"
export LINXA_TOKEN=YOUR_TOKEN
Install the skill:
clawhub install smart-linkedin-inbox
All requests require the LINXA_TOKEN environment variable. This is a secure bearer token — Linxa never asks for your LinkedIn password or session cookies.
Authorization: Bearer $LINXA_TOKEN
If the token is missing or expired, guide the user to regenerate it at app.uselinxa.com/setup-mcp.
Security model:
https://app.uselinxa.com
GET /api/mcp/current-li-user
Verifies authentication and returns the current LinkedIn profile. Call this first at the start of a session.
GET /api/mcp/conversations
Query parameters (all optional):
| Parameter | Type | Default | Description |
|---|---|---|---|
| ----------- | ------ | --------- | ------------- |
limit | integer | 50 | Max conversations to return |
search | string | — | Keyword search across messages and participants |
label | string | — | Filter by category label |
sentiment | string | — | POSITIVE, NEGATIVE, or NEUTRAL |
primary_intent | string | — | Filter by intent (e.g., "sales", "recruitment") |
intent_direction | string | — | to_me or from_me |
product | string | — | Filter by detected product interest |
Available labels: Hot, Need Follow Up, Personal, Investors, Clients, Inbox, Hiring, Junk, Partnership, archived, scheduled, not-contacted
GET /api/mcp/messages/{chatId}
Returns all messages in a specific conversation thread. The chatId comes from the conversation list response. URL-encode the chatId if it contains special characters.
POST /api/mcp/next-actions
Returns an AI-generated summary of recommended next actions across your LinkedIn conversations. Use this when the user asks "what should I do next?" or "what are my priorities on LinkedIn?"
POST /api/mcp/comments
Attach a CRM-style note to a LinkedIn lead's profile. Comments influence future next-action recommendations. Request body:
{
"profileId": "PROFILE_ID",
"text": "Follow up after demo on Friday"
}
POST /api/mcp/conversations/{chatId}/read
Marks a specific conversation as read. Use when the user says "mark this as read" or wants to clean up their inbox.
Use the helper script for authenticated requests:
bash scripts/linxa_api.sh GET /api/mcp/current-li-user
bash scripts/linxa_api.sh GET "/api/mcp/conversations?label=Hot&limit=5"
bash scripts/linxa_api.sh GET "/api/mcp/messages/CHAT_ID_HERE"
bash scripts/linxa_api.sh POST /api/mcp/next-actions
bash scripts/linxa_api.sh POST /api/mcp/comments '{"profileId":"PROFILE_ID","text":"Note here"}'
bash scripts/linxa_api.sh POST "/api/mcp/conversations/CHAT_ID/read"
Or curl directly:
curl -sL \
-H "Authorization: Bearer $LINXA_TOKEN" \
"https://app.uselinxa.com/api/mcp/conversations?limit=10&sentiment=POSITIVE"
/api/mcp/current-li-user to confirm the token workschatId from step 2 and pull full messagesWhen presenting conversations to the user:
| Problem | Solution |
|---|---|
| --------- | ---------- |
| 401 Unauthorized | Token is missing or expired — regenerate at Linxa dashboard |
| Empty results | Chrome extension may not be syncing — check extension is active and LinkedIn tab is open |
| chatId encoding errors | URL-encode the chatId value before making the request |
| No conversations found | Ensure you have LinkedIn conversations and the extension has synced recently |
For the complete OpenAPI specification, read references/openapi.yaml.
共 1 个版本