Workflow library
LinkedIn workflows for your AI assistant.
LinkedBoost exposes individual LinkedIn operations through a hosted MCP server. Start with one job on your agency's business-development account: research prospects, prepare a sequence, or work an existing conversation. Claude, ChatGPT, Instinct, and Meta Muse connections are confirmed working.
Use our agency's service brief to find ten prospects in our target market. Show fit evidence and prior contact before drafting. Do not enroll or send.
Profile research
Prospect shortlist
- Start with
- You have a role, market, and account you want to research.
- Workflow
- Search from your connected seat, inspect the returned people, and save a small lead group.
- Documented tools
- search_linkedin · continue_search · save_search
- Check the result
- Verify search entitlement and result freshness in your own account.
- Useful output
- A small saved group with a reason to include each person and a list of uncertain facts to check.
CSV to reviewed sequence
- Start with
- You have an authorized list of 10 to 25 people.
- Workflow
- Stage the CSV, save reviewed rows, inspect prior interactions, then prepare a sequence and review drafts before enrollment.
- Documented tools
- import_leads · save_search · get_person_interaction_history · create_sequence · get_sequence
- Check the result
- Confirm the selected account, recipient, limits, and final send state.
- Useful output
- A reviewed audience and sequence plan. Approve exact person IDs separately before enrollment queues outreach.
Inbox triage
- Start with
- You have an existing conversation that needs a response.
- Workflow
- Open the thread, inspect the context, draft a reply, and decide whether to send it.
- Documented tools
- list_conversations · get_conversation · send_reply
- Check the result
- Verify the reply appears on the correct thread after the action completes.
- Useful output
- A draft that answers the latest question, plus a decision to reply, pause, or leave the thread with its current owner.
Known-post engagement
- Start with
- You have identified a specific relevant post.
- Workflow
- Provide the post URL and verified content, inspect prior interactions, and prepare one useful comment or reaction for review.
- Documented tools
- get_person_interaction_history · comment_on_post · react_to_recent_post
- Check the result
- For a comment, supply the exact post URL and inspect the published result. The reaction tool targets a person's latest reactable post, which may change; skip it if you need one exact post.
- Useful output
- One relevant response using your actual experience. Skip the action if post context or a useful contribution is missing.
Keep a clear owner for the next step
Before outreach, identify the sender, audience, approved claims, and person handling replies. Check person and company history against your outside records. The operator resolves overlap; a missing record does not prove there has been no prior contact.
After approval, distinguish a saved draft, an enrolled person, a queued action, and a completed send. Inspect an ambiguous result before retrying. When someone replies or asks to stop, review remaining sequence steps before any further outreach.
What automation means here
An MCP-capable assistant can select and call documented tools during a session. Sequences have cadence and queued state. The operator remains responsible for reviewing actions, checking account limits, and verifying final outcomes.
What is not in this build
Continuous monitoring for every new post, unattended auto-commenting, new-message triggers with autonomous replies, original-post publishing, native CRM or email sync, client dashboards, and white-label reports are not offered here.
The LinkedIn User Agreement prohibits unauthorized automation. Product limits and review do not eliminate account risk. Start with an account you are authorized to operate and review the policy before outreach.