TT Claude Memory

feedback_always_display_full_scorecard

Properties1
description2026-04-23 Rich explicit: "please display the full scorecard - i will always need to see the scorecard in full, for every question - regardless of how tight the contest between the answers is". Summarising a saved scorecard and pointing to the file is not enough — the full weighted table must be rendered inline in every scorecard turn.

3 min read

Whenever a weighted scorecard is produced for an INHERIT v2 / TT-strategy decision, display the full scorecard inline in the chat turn where the decision is being presented, even after saving it to disk.

Why

Rich’s explicit directive on 2026-04-23 during Phase 5 adjudication of the INHERIT v2 technical audit:

“please display the full scorecard - i will always need to see the scorecard in full, for every question - regardless of how tight the contest between the answers is”

Pointing Rich at a saved file (docs/superpowers/scoping/…/T-XXX-scorecard.md) and summarising the verdict is INSUFFICIENT. Rich reads the full weighted table in-chat when adjudicating — the summary-plus-link pattern forces him to switch contexts and re-read, which slows adjudication and makes comparison across scorecards harder within a single session.

This rule compounds with feedback_always_save_scorecards (2026-04-24): every scorecard is both saved to disk AND displayed in full in chat. Not one-or-the-other.

How to apply

Every time a scorecard is produced:

  1. Save to file per feedback_always_save_scorecards discipline (required location + frontmatter + structure)
  2. Then display the full scorecard in chat, including:
    • Question (restated)
    • Options (all, with brief descriptions)
    • Weights table (15 criteria + weight + rationale for weight)
    • Scoring matrix (15 × N options, weighted totals + percentages)
    • Verdict (winner + margin)
    • Sensitivity analysis (at least 2-3 stress tests)
    • Honest critique
    • Locked-in cascade effects
  3. Then invite Rich to confirm / override / re-weight / reframe

The display can be compact (table formatting in markdown renders cleanly in Claude Code) but must be complete. Do not abbreviate the scoring matrix. Do not hide the weights table behind “see file for details”. Do not elide the sensitivity analysis.

What NOT to do

  • ❌ “Full scorecard saved at T-XXX-scorecard.md. Verdict: Option D at 89%. Ready to move on?” — this is the anti-pattern Rich corrected.
  • ❌ Display only the verdict + top-line percentages without the weighted-scoring matrix
  • ❌ Display only the weighted totals without the 15-criterion breakdown
  • ❌ Omit the sensitivity analysis “because the contest wasn’t tight” — Rich wants it regardless of margin
  • ❌ “The scorecard is long — want me to paste it?” — just paste it; don’t ask permission

What TO do

  • ✅ Render the full saved scorecard content directly in the chat turn
  • ✅ Use markdown tables (they render cleanly); prefer compact cell content (2-6 words per cell) to keep readable width
  • ✅ Include every section the saved file includes: question / options / weights / scoring / verdict / sensitivity / critique / locked-in
  • ✅ Offer to re-weight or reframe at the end of the display — don’t wait for Rich to request it

Triggers

Any scorecard for:

  • Audit-tension adjudication (Phase 5 of the INHERIT v2 technical audit)
  • Shape X angle decisions
  • Architecture-option choices
  • Amendment-numbering or module-boundary decisions
  • Any other weighted-decision artifact under docs/superpowers/scoping/*-scorecards/
  • feedback_always_save_scorecards.md (2026-04-24) — save every scorecard to disk; do not rely on in-chat-only tables
  • feedback_reframe_beats_reweight.md — if a scorecard stalls at 85-92%, question framing before re-weighting
  • feedback_strategy_conversations.md — Rich prefers tiered framings + honest self-correction

Graph View

Backlinks