Frontend
Undo & Optimistic Actions
Implements reversible actions and optimistic UI that stays honest about what the server actually confirmed.
What it does
Implements reversible actions and optimistic UI that stays honest about what the server actually confirmed.
When to use it
Use for undo-after-delete, optimistic create/update, or audits where fast UI feedback conceals a failed operation.
Problems it prevents
Stops an optimistic UI update sticking around after the server has actually rejected the change.
Typical workflow
- Choose the right protection (undo vs confirm)
- Model the mutation and rollback path
- Implement the undo window
- Reconcile with authoritative server state
Example output
A delete action with a real undo window that only becomes permanent once the server has actually confirmed it.
What's included
- 01The complete prompt
The full manual as a ready-to-save SKILL.md file, matched exactly to what ships in the T7 library.
- 02Install walkthrough
Personal (~/.claude/skills/) and project (.claude/skills/) folder instructions, whichever fits how you work.
- 03Use notes
How Claude invokes it, what a good result looks like, and what to check before you trust the output.
Installation
Save the manual as SKILL.md in ~/.claude/skills/undo-optimistic-actions/ to use it on every project, or in .claude/skills/undo-optimistic-actions/ to scope it to one repository. Then invoke it directly:
/undo-optimistic-actionsWorks well with
Recommended bundle
Save by bundling
Get Undo & Optimistic Actions as part of Reliability Builder: five skills that work together as one toolkit, for £1.96 less than buying all five separately.
