Keep a decision journal with AI
Record what you knew, what you chose and when to revisit it, then compare the original reasoning with the outcome.
Keep a decision journal with AI when a choice will be easy to reinterpret after the results come in. Extending a shop’s opening hours might be sensible given the evidence available today. The decision-journal skill helps an agent record that evidence, the alternatives and the reason for the final choice while those details are still fresh.
A review can then separate process from luck. An unforeseen closure could spoil a well-reasoned choice, while a fortunate sale could flatter a poorly examined one. Appending a dated outcome preserves both accounts and makes the next lesson easier to use.
What it does
- Write down the options that were genuinely available.
- Record the reasoning and the evidence known at the time.
- Identify two specific ways the choice could go wrong.
- Set a review date connected to an observable outcome.
- Append the result and one lesson for a future decision.
Install and use
- Give the agent the decision, options and a realistic review date. Keeping the record locally requires an authorised folder and a writable Markdown journal.
- Ask, “Record why we chose Saturday opening over longer weekdays. Name two ways it could go badly and set a review after three weeks.”
- At review time, compare what happened with the original account before writing the dated outcome.
FAQ
Source
Made by Gen AI Hub (Megan Martinsone). Original skill: decision-journal, published on GitHub under the MIT licence.
View on GitHub