Compare Two Versions of a Prompt
See exactly what changed between v1 and v2 of a prompt — added, removed, and modified instructions, plus whether the revision reduced or introduced risk.
When a prompt's output suddenly shifts, the first question is what changed in the prompt. A diff answers it in seconds — additions, removals, and edits, itemized.
The usual debugging story: a prompt that produced reliable output starts behaving differently, and the only suspect is an edit someone made along the way. Re-reading the prompt won't surface the change — you remember the intent, not the wording. Diffing the current version against the last known-good one shows the exact instructions that were added, removed, or reworded, and flags whether any of them moved a risk signal. The loaded pair reproduces this scenario: a small 'harmless' edit that actually dropped a constraint.
Diff the loaded pair
Version B looks like a harmless extension — it adds milestones with dates. Run the diff.
Find the silent removal
The 'do not mention internal discussions or unconfirmed timelines' constraint is gone — and B now asks for dates, which that constraint used to guard.
Read the risk changes
Removed constraints register as an output-control decrease. That's the regression the edit's author never noticed.
Apply to your incident
Paste the last known-good version as A and the current one as B — the suspect edit is in the removed or modified list.
It itemizes additions, removals, and edits, so a dropped line surfaces in the removed list even when the edit looked like an extension. In the loaded pair, version B reads as a harmless add (milestones with dates) but quietly deleted "do not mention internal discussions or unconfirmed timelines" — the constraint that used to guard the very dates B now requests.
Paste the last known-good version as A and the current one as B, per the workflow's final step; the suspect edit lands in the removed or modified list. The diff recalls the wording your memory doesn't, and flags removed constraints as an output-control decrease — but it compares two prompts, it doesn't run them or confirm the fix.
See exactly what changed between v1 and v2 of a prompt — added, removed, and modified instructions, plus whether the revision reduced or introduced risk.
A lightweight way to track how a prompt changes over time: keep the previous version, diff every revision, and read the risk deltas instead of guessing.
A worked marketing prompt revision — from adjective soup to offer-driven — with the diff showing exactly which changes carry the improvement.
A side-by-side way to decide between two ChatGPT prompt drafts — scored on clarity, specificity, output control, and risk instead of gut feeling.
A set of before-and-after examples showing exactly what prompt cleanup removes — and what it deliberately leaves alone.
Formats fuzzy agent instructions into a structured prompt with objective, available tools, constraints, success criteria, and failure handling.
Diff two versions of the same prompt — added, removed, and modified instructions, plus risk changes.