- Who it helps
- People considering a review of their own coding-assistant chats.
- Availability
- Public desktop project; requires a chosen model provider and its credentials. Provider charges can apply.
What leaves the computer
The README describes sending selected excerpts to the provider after consent. It removes obvious credentials, but personal text can remain. Analysis therefore involves a remote provider.
Recursive Human Intelligence reviews the prompts in your past chats ↗
Check the background setting
After connection and consent, a scheduled Sunday run can send excerpts without another prompt. Disable Background in Settings if you want manual reviews. Excluding a source stops future imports; deleting retained history is a separate action.
Recursive Human Intelligence reviews the prompts in your past chats ↗
What you can do
- Read the data and background-run settings before connecting a provider. Open reference ↗
Where the evidence stops
README checked November 3. No installation, private-log access or data-flow test. The repository’s demonstration is scripted.
Sources read. Sources read November 3, 2026 UTC. No third-party software or complete workflow executed.
Sources & further reading
GitHub · implementation
Recursive Human Intelligence reviews the prompts in your past chats ↗
Identifies the subject, documented capability or specific verification limit.
Andrea Ritossa · page checked 2026-11-03
Coverage note: Independent attention and end-to-end execution remain unverified.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Added to Current
Recursive Human Intelligence turns past assistant conversations into advice about how you ask for help.