Skip to main content
Glama
antoineDsh

SalesTouch

SalesTouch

Version 0.9.3 License MIT

LinkedIn outreach for AI agents. SalesTouch is a LinkedIn MCP connecting Claude, Codex, ChatGPT, Cursor, and other MCP-compatible agents to LinkedIn search, research, conversations, messaging, engagement, publishing, durable scheduling, queues, account controls, and paginated extraction through an OAuth-protected remote endpoint.

SalesTouch is not affiliated with, endorsed by, or sponsored by LinkedIn.

Install

Claude Desktop, Cowork, or Claude Code

See the Claude installation guide, available in English, French, German, Spanish, Brazilian Portuguese and Italian.

In Claude Code:

/plugin marketplace add antoinedsh/salestouch
/plugin install salestouch@salestouch
/mcp

Cursor

For local plugin installation, clone this public repository into ~/.cursor/plugins/local/salestouch, then reload Cursor. Open Settings → Tools & MCPs, find salestouch, click Connect, and complete the SalesTouch OAuth flow in your browser. Local plugin availability depends on organization policy. Marketplace listing remains subject to Cursor review.

Alternatively, add this remote MCP configuration to .cursor/mcp.json in your project or ~/.cursor/mcp.json for your user profile:

{
  "mcpServers": {
    "salestouch": {
      "url": "https://www.salestouch.io/api/mcp"
    }
  }
}

Authorize promptly: Cursor may stop waiting after 30 seconds. If that happens, click Connect again and complete the new authorization flow. Once connected, use the read-only account prompt below. Keep tool approvals enabled and review recipients and content before allowing external actions.

The MIT plugin is free to install. The hosted SalesTouch service requires a separate subscription. No LinkedIn credentials belong in the MCP configuration.

Français · Deutsch · Español · Português · Italiano

Any Streamable HTTP client

Use https://www.salestouch.io/api/mcp and name the server salestouch. The client discovers the SalesTouch OAuth flow automatically; no static API key or LinkedIn credential belongs in the MCP configuration.

Related MCP server: linkedin-automation-mcp

OAuth and first check

The browser authorization flow grants scoped SalesTouch access. LinkedIn passwords, cookies, and authentication secrets are never shared with the AI client.

Verify the connection with a read-only request:

List the LinkedIn accounts connected to my SalesTouch workspace.

Example workflows

1. Research a lead without changing external state

  1. Ask SalesTouch to look up a profile URL.

  2. Ask for the profile's recent posts, then continue with next_cursor when older history is needed.

  3. Ask the agent to summarize useful context and draft a message.

  4. Review the draft; no message is sent until a mutation tool is authorized.

2. Review a conversation and send an exact reply

  1. Search the conversation by profile URL.

  2. Read the bounded message history.

  3. Ask the agent to draft a reply without sending.

  4. Confirm the recipient and exact text, then authorize linkedin_message.

3. Build a Sales Navigator audience from structured filters

  1. Resolve ambiguous filter values with linkedin_sales_navigator_search_parameters.

  2. Run scrape_sales_navigator_companies or scrape_sales_navigator_people with the returned ids.

  3. To restrict people to extracted companies, pass the company result_id to scrape_sales_navigator_people as company_result_id.

  4. Inspect the bounded preview, page with scrape_result_page, or use the signed CSV/JSONL export.

4. Turn post engagement into a paginated result

  1. Provide a LinkedIn post URL.

  2. Run scrape_linkedin_post_reactions or scrape_linkedin_post_comments. Comment scraping extracts root commenters by default; pass include_replies: true only when reply authors are needed.

  3. Read the bounded preview and stored result id.

  4. If a comment result returns source_paging.exhausted: false, call scrape_linkedin_post_comments again with source_paging.arguments to continue the LinkedIn source extraction in the same reply mode. Large sources pause automatically after a bounded provider-request budget, so a call may return fewer profiles than limit without losing progress.

  5. Use scrape_result_page only to page through profiles already stored by one scrape call.

Safety and limitations

  • Read tools are non-destructive and idempotent.

  • Messages, invitations, likes, comments, and publications modify LinkedIn and are non-idempotent.

  • These write tools accept scheduled_for; SalesTouch persists the action and reconnects a sleeping credential-based account when execution becomes due.

  • Cancelling a queued action modifies SalesTouch state and cannot be assumed idempotent.

  • Availability depends on an active SalesTouch plan, a connected LinkedIn account, platform availability, permissions, and usage limits.

  • SalesTouch does not bypass LinkedIn safeguards. Users are responsible for reviewing targets and content and for complying with applicable platform rules and law.

Troubleshooting

If OAuth fails, reconnect from the MCP client, confirm the endpoint, verify the SalesTouch plan and connected account, then retry the read-only account check. Never paste credentials, cookies, OAuth tokens, private messages, or prospect data into a support request.

Documentation and support

The files in this public distribution wrapper are licensed under the MIT License. The hosted SalesTouch backend remains proprietary.

Related MCP Connectors

Related MCP Servers

  • A
    license
    A
    quality
    D
    maintenance
    Fully featured MCP server that provides automation tools for LinkedIn, supporting browser-based scraping and API-based operations for content management, media uploads, and reactions.
    6
    3
    MIT
  • F
    license
    Not graded
    quality
    C
    maintenance
    MCP server for LinkedIn API integration. Enables authentication, profile access, connections, search, messaging, and feed management via OAuth2.
    -
  • A
    license
    Not graded
    quality
    B
    maintenance
    A production-ready MCP server that enables AI agents to autonomously create, manage, and engage with LinkedIn content through the official LinkedIn REST API.
    124 npm
    2
    MIT