← all meta proposals

cheapest_instant_kill_test extraction harness from council_verdict

filter rejected HARNESS reversible: simple 4h proposed 1 Jun 2026
What is the proposed change?
council_verdict.js:26-43 already requires every judge to emit a cheapest_instant_kill_test field. Today the field is logged inside the move's output blob and never surfaced anywhere downstream. Add: after a council round resolves (any action other than 3-0 KILL), collect the cheapest_instant_kill_test strings from all 3 judges' parsed verdicts, dedupe by trigram overlap >70%, and write the resulting array of 1-3 tests into a new key 'kill_test_registry' inside the existing hypothesis dossier_json column (no schema migration — dossier_json is already a free-form blob field). brief.js:102 currently surfaces median + IQR in the markdown brief; append a 'Cheapest instant kill tests:' bullet list rendered from kill_test_registry. Each entry tagged with the originating judge model and verdict_action.
Target files
hypothesis_engine/moves/council_verdict.js hypothesis_engine/brief.js
Expected effect
Every active non-killed hypothesis after first council pass will have 1-3 kill tests surfaced in the brief Commander reads. Over a 14-day window, the brief becomes more actionable: Commander can scan kill tests and run the cheapest one rather than re-reading 9 deep reports.
Falsifier — what would prove this wrong?
After 14 days, audit kill_test_registry entries across all hypotheses that went through council. (a) If <80% of entries are unique strings (lots of generic 'email 10 prospects' duplicates), the trigram dedupe is too loose OR the judges emit generic tests — harness is recording noise. (b) If Commander reports during the 14-day window that no kill test in any registry was actually run, the harness surfaces text Commander does not use, and the brief.js render is the wrong audience.
Evidence that triggered the proposal
  • D — hypothesis_engine/moves/council_verdict.js:26-43 — cheapest_instant_kill_test field already in judge schema, currently unused downstream
  • D — hypothesis_engine/brief.js:102 — brief render path currently only shows filter score median + IQR

Proposer self-score

The proposer scored its own draft on these axes (0-3 each) before submitting.

AxisScore
specificity3
falsifier2
solo feasible3
blast radius3
composability3
reversibility3
Disposition
Rejected by filter_score. The proposal did not meet the bar for specificity, falsifiability, or solo-feasibility.

Evaluation history

WhenMove
2026-06-01 04:08meta_filter_score
2026-06-01 04:05meta_genesis