| name | outcome-memory |
| description | Close the loop between recommendations and real outcomes so future Codex judgment improves over time. |
Outcome Memory
Use this skill after a plan, review, ship, design decision, startup memo, or ethics review has had enough time to produce a result.
This is a Praneet-specific hand-port enhancement. It makes judgment compound by recording whether prior advice worked, failed, or needs revision.
Workflow
- Identify the prior decision or recommendation.
- Gather outcome evidence:
- what happened
- what shipped
- user or stakeholder response
- metrics or qualitative signal
- surprises
- harms or exclusions discovered later
- Compare expectation to reality:
- right for the right reason
- right for the wrong reason
- wrong but useful
- wrong and misleading
- still unknown
- Extract learning:
- update future scorecards
- revise assumptions
- add follow-up tasks
- capture ethics or accessibility lessons
- Persist:
- write a report when the outcome should be reviewable
- use
capture or brain-ops for durable memory
- route to
retro when this is team/process learning
Guardrails
- Do not force a lesson when evidence is still ambiguous.
- Do not hide negative outcomes.
- Separate personal preference drift from actual user or ethical evidence.
