LinkedIn MCP prompts for Claude, ChatGPT and Codex
Paste these into any AI client connected to the Zevari LinkedIn MCP. Each one hands a job to one of the six specialists: Prospecting, Signals, ICP Scoring, Voice Writer, Follow-up and Reply Qualifier. Every send still waits for your approval by default.
How Claude, ChatGPT and Codex works with Zevari
- Signals: find a reason to reach out
Find people who posted about hiring their first SDR in the last 30 days. Rank them by how strong the signal is.
- Voice Writer: draft a connection note
Draft three short LinkedIn connection notes for this prospect in my voice. Reference their recent post and do not pitch. Stage the best one for my approval.
- Reply Qualifier: triage the inbox
Go through this week's LinkedIn replies. Show me the qualified conversations first, with the next step for each, and draft replies for my approval.
Common workflows
- Research prospects before outreach
Ask your AI client to research people, companies, triggers, and context before writing the first message.
- Draft LinkedIn connection notes, DMs, and content
Generate channel-aware outreach and content that matches your positioning, ICP, tone, and the prospect context.
- Stage actions for review
Prepare connection requests, replies, comments, follow-ups, content, and campaign actions in Zevari before sensitive actions execute.
- Classify replies and inbox intent
Turn inbound conversations into next steps, hot leads, follow-up tasks, and pipeline updates.
Safety controls
- Confirmation gates
Sensitive write actions, including messages, comments, connection requests, content publishing, and campaign activation, are routed through review boundaries.
- Account-level controls
Zevari applies pacing, duplicate prevention, stale-state checks, and account-state checks before risky GTM actions proceed.
- Reviewable workspace state
Operators can inspect staged work, recent activity, blocked actions, pending approvals, signals, campaigns, and inbox context in Zevari.