Andrew Reed c9ce77a70f Link CI eval results to PRs and speed up vectorstore build via gateway ...

- Write a GitHub Job Summary with a direct link to the LangSmith
  experiment, so the PR check is one click from the actual results.
- Trace experiment names/metadata back to the originating PR and commit
  (ci-pr<N>-<sha>-<hash>) instead of a generic ci-regression-<hash>.
- Route the CI vectorstore build through LangSmith's LLM Gateway with
  OpenAI embeddings (auth'd via the existing LANGSMITH_API_KEY secret,
  no code changes needed) instead of local HuggingFace, cutting the
  build step from ~60s+ to ~6s.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

2026-07-24 09:46:53 -04:00

..

section_1_production_data_flywheel.ipynb

update new repo/project name throughout

2025-11-25 10:51:28 -05:00

section_2_sdk_interaction.ipynb

updates

2026-01-06 21:56:59 -05:00

section_3_cicd_regression_gate.ipynb

Link CI eval results to PRs and speed up vectorstore build via gateway

2026-07-24 09:46:53 -04:00