John SmithandClaude de7d011175 fix(llm+sla): empty-stream honesty; SLA rows disclose what they measure (iteration 70) ...

Three small audit notes closed red-first:
- Empty/[DONE]-only SSE reassembly no longer masquerades as a
  completed turn: complete_with_tools raises LLM_UPSTREAM_EMPTY and the
  tool agent surfaces that status instead of COMPLETED with blank
  narration.
- The SLA 'intent_parsing' row is relabelled intent_parsing_keyword
  (it measures regex classification); an LLM-classification row appears
  only under RUN_LIVE_LLM=1 with a real key, counting only genuine llm
  outcomes; SSE first-token requires >=3 samples or WITHHELD.
- Report ends with explicit FAILED/PASSED/WITHHELD tallies and an
  all-withheld run exits nonzero — withheld can never read green.
21 red-first tests; suite at 451.

Co-Authored-By: Claude <[email protected]>

2026-08-24 08:25:37 -04:00

..

tests

fix(llm+sla): empty-stream honesty; SLA rows disclose what they measure (iteration 70)

2026-08-24 08:25:37 -04:00

vtt_orchestrator

fix(llm+sla): empty-stream honesty; SLA rows disclose what they measure (iteration 70)

2026-08-24 08:25:37 -04:00

pyproject.toml

test(llm): opt-in live-LLM suite that never runs in CI (iteration 55)

2026-08-24 06:24:12 -04:00