Overview
How might we design ai command history so people can trust and act on AI output?
When to use
- Ideal for power users, repetitive workflows, and applications where quick access to and replay of past AI commands improves productivity.
When to skip
- Shared devices where history exposes private prompts.
- Retention policies requiring no local command storage.
- History with no way to delete sensitive entries.
Rules
History that stores secrets in plain text with no redact.
Replay that reruns destructive tools without confirm.
Unbounded list with no search.
History separate from the composer with no insert action.
Evidence
| Product | Implementation |
|---|---|
| Terminal history | Up-arrow recall with query search in shells. |
| Command palettes | Recent commands section with rerun. |
| AI chat history | Thread list with fork and continue actions. |
| Automation tools | Run logs with edit-and-retry on failed steps. |