Commits

Commit history

1. Commits on Oct 28, 2026

   • add human review of 10 held-out answers by the project owner; recorded in report, system card and README

     krahman

     committed
     f05a2bdView commit details

     Copy full SHA for f05a2bd

     View code at this pointBrowse repository at this point

   • added human review sheet for 10 held-out answers (blank)

     krahman

     committed
     bba580eView commit details

     Copy full SHA for bba580e

     View code at this pointBrowse repository at this point
2. Commits on Oct 27, 2026

   • final simplified agent, held-out results, report rewritten to the spec, removed empty mcp_server folder

     krahman

     committed
     cc083e5View commit details

     Copy full SHA for cc083e5

     View code at this pointBrowse repository at this point

   • simplified the agent (no router), removed extras beyond the spec, rescored dev results

     krahman

     committed
     0eeec80View commit details

     Copy full SHA for 0eeec80

     View code at this pointBrowse repository at this point

   • added --no-upload to the CLI and docs/OPEN_ITEMS.md

     krahman

     committed
     d5f0cddView commit details

     Copy full SHA for d5f0cdd

     View code at this pointBrowse repository at this point

   • final evaluation on held-out split, red team, human review sheet

     krahman

     committed
     47c2121View commit details

     Copy full SHA for 47c2121

     View code at this pointBrowse repository at this point

   • froze final configuration; added budget guard, red-team acceptance rules, final analysis

     krahman

     committed
     dd7b9a2View commit details

     Copy full SHA for dd7b9a2

     View code at this pointBrowse repository at this point

   • added red-team pass, output tripwire, resource tool, guardrail tests, system card

     krahman

     committed
     a02b108View commit details

     Copy full SHA for a02b108

     View code at this pointBrowse repository at this point

   • corrected F05 and K02 labels (v1.1) and re-scored E5-E7 outputs

     krahman

     committed
     cf97177View commit details

     Copy full SHA for cf97177

     View code at this pointBrowse repository at this point

   • added agent comparison (E7), H3 analysis and demo results

     krahman

     committed
     3259be5View commit details

     Copy full SHA for 3259be5

     View code at this pointBrowse repository at this point

   • added MCP server, validating client, protocol demo

     krahman

     committed
     2d45ae0View commit details

     Copy full SHA for 2d45ae0

     View code at this pointBrowse repository at this point

   • added generation chain, validation V1-V6, LLM judges, E5/E6

     krahman

     committed
     4eb91a8View commit details

     Copy full SHA for 4eb91a8

     View code at this pointBrowse repository at this point

   • refreshed retrieval results with cold caches

     krahman

     committed
     8556a2bView commit details

     Copy full SHA for 8556a2b

     View code at this pointBrowse repository at this point

   • added retrieval experiments (hybrid, re-rank, period filter) with bootstrap CIs

     krahman

     committed
     3f88dc2View commit details

     Copy full SHA for 3f88dc2

     View code at this pointBrowse repository at this point

   • re-ran baselines on clean commit

     krahman

     committed
     800cb87View commit details

     Copy full SHA for 800cb87

     View code at this pointBrowse repository at this point

   • added baselines (BM25, dense RAG, rule-based) and evaluators

     krahman

     committed
     8bad045View commit details

     Copy full SHA for 8bad045

     View code at this pointBrowse repository at this point

   • added evaluation set

     krahman

     committed
     7edc331View commit details

     Copy full SHA for 7edc331

     View code at this pointBrowse repository at this point

   • added project scaffolding

     krahman

     committed
     6dc3c8eView commit details

     Copy full SHA for 6dc3c8e

     View code at this pointBrowse repository at this point
3. Loading