Record work decisions with AI
Preserve a decision’s reasoning, rejected alternatives and a checkable condition for revisiting it in a new, structured Markdown note.
Record work decisions with AI while the reasoning is still available. The remember skill focuses on one choice and asks a short set of questions about what you rejected, the constraint that settled the issue and what would invalidate the decision.
The result is a structured Markdown note rather than a tidy conclusion stripped of its context. Portable lessons stay separate from employer-specific evidence. Existing memory notes are not edited: a later correction creates a new record that explicitly supersedes the earlier one.
What it does
- Keep the decision in your own words
- Record rejected alternatives and their reasons
- Define a checkable condition for reconsideration
- Separate portable lessons from proprietary evidence
- Link supporting material without copying whole logs
- Create a traceable successor when a decision changes
Install and use
- Load the self-contained instruction file and licence in a file-writing agent and choose a memory directory. Plain local writing is sufficient; PowerShell and Git backup are separate optional setup.
- Ask: “Record why we chose flat files over a database for internal reports. Ask me about the alternatives we rejected.”
- Answer the missing questions and confirm the scope. Keep unrelated private material and credentials out, and retain your own wording for the decision and deciding constraint.
- Review evidence and the invalidation condition. Unanswered questions produce an inbox note, not an approved memory; promotion is a human action.
FAQ
Source
Made by Mason Tseng. Original skill: remember, published on GitHub under the MIT licence.
View on GitHub