Primary skill

`tcx-memory`

Retrieve prior decisions, replay historical judgment with point-in-time evidence, review outcomes, and validate whether a lesson actually holds.

Ask a historical question #

Use this skill when you need to understand what was known at an earlier point, revisit a thesis change, compare process quality with outcome quality, or find supporting and contrary past episodes.

Example

$tcx-memory
Review the outcome of my latest NVDA thesis. Separate decision quality from outcome quality.

Replay without hindsight #

A replay uses point-in-time source snapshots and decision records rather than a later-updated summary. For current judgment, the system forms and records an independent view from current evidence before it retrieves similar past cases; then it records how memory changed or did not change that view.

Outputs can include source-bound episodes, replay/review artifacts, lesson status, evidence tier, and the next validation required.

Keep memory non-authoritative #

Memory is evidence and judgment support, not authority. It cannot authorize a recommendation, order, approval, broker action, execution, policy change, or automatic self-modification.

A validated lesson can become an input to a separate Brain authoring task, but memory does not install or modify a Brain or Strategy by itself. See Decision Memory for the full model.