Audit your posts for promises you never kept

They used Claude to scan recent posts for commitments and uncovered a false alarm caused by test content.

Done with Claudeยท Advancedยท Sep 2026

What they did
They built a Claude-made skill that calls list_posts, reviews their latest twenty posts, and looks for repeated topics and dangling threads. When it ran, it marked two LinkedIn posts from July as an unfulfilled Zapier promise. Those posts were only review tests, so they revised the skill and added filters plus a rule that it must ask before treating a guess as fact.
How it went
The updated version now gives three possible angles and stops at the drafting stage. Early misfires taught them that QA posts can look like real commitments, that outside references may be missing, and that a draft can still read badly even after passing a checker.
Tips
Exclude test and QA material before drawing conclusions from posting history. Have the agent confirm any history-based claim before you rely on it.

Try it yourself with Claude Code

Review my posting history from [platform] at [file path]. Look for places where I seem to have made a commitment or set up a future follow-up, then check whether I actually followed through. List each item with the date, what was implied, and whether there is clear evidence it was completed. If something is unclear or depends on an assumption, ask me before treating it as true. Do not send, publish, delete, or submit anything; stop and show me a draft for approval first.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on dev.to โ†—

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.