Blog

How to Automate LinkedIn Research and Drafting with Claude Code and Wonda

By Wonda Teamtutorials
How to Automate LinkedIn Research and Drafting with Claude Code and Wonda hero image
Use Claude Code and Wonda to research LinkedIn conversations, study what peers are posting, draft better posts, and stay human on the final publish step.

LinkedIn is one of the easiest places to look fake.

That is why the only automation workflow worth keeping is the one that helps you think better before you post.

Claude Code is useful here because the real LinkedIn job is not typing into the composer. It is researching the conversation around your niche, reading what peers are posting, noticing what angles are tired, and drafting something that still sounds like a person. Wonda gives Claude Code the LinkedIn commands to actually do that work from the terminal.

This guide focuses on the workflow we think is defensible: automate research, summarization, draft generation, and inbox triage. Keep the final judgment with the person whose name is on the post.

Key Takeaways

  • LinkedIn automation is most useful before the publish button, not after it.
  • Claude Code can turn LinkedIn research into better post ideas and better first drafts.
  • Wonda gives Claude Code access to search, profiles, posts, notifications, conversations, and publishing.
  • Human review matters more on LinkedIn than most people admit.

Why LinkedIn Needs a Different Kind of Automation

Most LinkedIn tools optimize for scheduling.

That is not useless, but it is not the hard part.

The hard part is writing something that:

  • is specific enough to sound real
  • is relevant enough to get read
  • is shaped enough to work on LinkedIn
  • does not sound like it was generated by "B2B thought leadership machine v4"

That is why Claude Code helps more with research and drafting than with scheduling.

If you feed it:

  • a handful of strong profiles
  • a few recent posts
  • your own product context
  • the tone you want to keep

it can produce a much better first draft than a blank LinkedIn composer ever will.

What Wonda Gives Claude Code on LinkedIn

Wonda exposes the useful parts of LinkedIn as commands:

  • wonda linkedin search
  • wonda linkedin profile
  • wonda linkedin posts
  • wonda linkedin company
  • wonda linkedin notifications
  • wonda linkedin conversations
  • wonda linkedin messages
  • wonda linkedin post
  • wonda linkedin like

That turns LinkedIn from a browser task into something Claude Code can inspect and operate from one session.

Setup

Install Wonda:

curl -fsSL https://wonda.sh/install.sh | bash
wonda auth login

Make sure the LinkedIn command surface is visible:

wonda linkedin --help

Then connect your session:

wonda linkedin auth set \
  --li-at-value "<li_at>" \
  --jsessionid-value "<JSESSIONID>"

wonda linkedin auth check

That gives Claude Code access to the same LinkedIn workflow you would otherwise manage manually.

Step 1: Research the People Already Shaping the Conversation

Start by asking Claude Code to map your niche.

Example prompt:

Use Wonda to find founders, operators, and marketers posting about
AI agents, social workflows, and creator automation on LinkedIn.
Show me the people and companies that come up repeatedly.

The command pattern behind that is usually:

wonda linkedin search "AI agents" --type PEOPLE -n 10
wonda linkedin search "social media automation" --type PEOPLE -n 10
wonda linkedin search "creator tools" --type COMPANIES -n 10

This gives you a faster version of the research most people do manually:

  • who is active in the niche
  • which companies keep appearing
  • what job titles cluster around the topic

Step 2: Study What the Good Posts Actually Look Like

Once Claude Code finds the right people, look at the recent posts:

wonda linkedin posts <vanity-name> -n 10

Then ask:

Read the recent posts from these profiles and tell me:
- which formats seem to get discussion
- which topics are already saturated
- what feels specific and believable
- what sounds like generic LinkedIn filler

This is where Claude Code is genuinely useful. It can quickly separate:

  • real operator posts
  • bland growth content
  • recycled hot takes
  • posts with a concrete angle worth responding to

That makes your own drafting much stronger.

Step 3: Turn Research Into Better Drafts

This is the core loop.

Instead of prompting:

"Write me a LinkedIn post about AI automation"

prompt something like:

Based on the profiles and posts we just reviewed, draft a LinkedIn post
about using AI agents for marketing workflows.
Make it sound like an operator note, not thought leadership.
Use first person. Keep it concrete. No hype words.

That gives Claude Code enough context to write something grounded.

If you want to tighten the voice further, add constraints:

Keep it under 220 words.
Use short paragraphs.
Avoid clichés like "game changer", "revolutionize", or "in today's landscape".

That is usually enough to get a draft worth editing.

Step 4: Publish Only After Review

Once you have a draft you actually like:

wonda linkedin post "..."

You can also set visibility:

wonda linkedin post "..." --visibility ANYONE

The important thing is not the command. It is the review step before it.

LinkedIn is attached to your name, your company, and your relationships. It is not a place to let an agent freestyle publicly without supervision.

Step 5: Use Claude Code for Notification Triage Too

LinkedIn automation is not just about creating posts. It is also about not losing the follow-up work after a post performs.

Use:

wonda linkedin notifications -n 20
wonda linkedin conversations

Then ask Claude Code:

Summarize what needs attention:
- comments worth replying to
- messages worth answering
- low-signal notifications I can ignore

This is a small workflow, but it is one of the most useful ones. It keeps you from confusing activity with actual things that need a response.

A Practical Weekly Workflow

If you are a founder or operator posting one to three times per week, this is enough:

  1. research 5-10 relevant profiles
  2. read 20-30 recent posts total
  3. identify what feels stale and what still feels sharp
  4. draft one post from a real observation
  5. edit it yourself
  6. publish
  7. use Claude Code to triage notifications for the next day or two

That is a much better loop than scheduling three low-conviction posts in advance.

What Still Needs Human Judgment

Voice

Claude Code can imitate a tone. It still cannot fully own your voice.

Relationship sensitivity

The right reply to a customer, investor, recruiter, or peer is not interchangeable.

Positioning

Not every trending topic is worth joining. Sometimes the best brand move is to stay out of the conversation.

FAQ

Is this mostly a post-scheduling workflow?

No. The value is in research, synthesis, draft generation, and notification review.

Can Claude Code write the full post?

Yes. But the best workflow is still draft first, human edit second.

Do I need LinkedIn's official API for this?

No. Wonda gives Claude Code a direct LinkedIn command surface for the workflow described here.

What is the best first use case?

Profile and topic research. It is the fastest way to improve what you post without sounding automated.

Final Advice

Good LinkedIn automation does not make you louder.

It makes you more deliberate.

If Claude Code and Wonda help you research faster, draft better, and respond to the right things, the workflow will feel natural. If they push you toward more generic output, stop and tighten the loop.

If you want the Codex version of the same idea, read How to Automate LinkedIn with OpenAI Codex CLI and Wonda.

#linkedin#claude-code#automation#cli#wonda