Repository navigation
feat(onboarding): metadata-filter vocab as per-client config + guided-review worksheet - #136
Merged
Merged
Conversation
…-review worksheet The lightweight per-client onboarding path (split from #127, which is the heavy RAGAS-benchmark add-on). Two gaps the client docs describe but don't provide. - Metadata-filter vocabulary moves from hardcoded module constants in query_pipeline.py to env-overridable JSON in pipeline/constants.py: METADATA_FILTER_DOCTYPE_KEYWORDS and METADATA_FILTER_STATUS_KEYWORDS. The status form is a {keyword: status-value} map (not a list + .capitalize()), so a client can point e.g. "expired" -> "Inactive". Defaults reproduce the shipped behaviour exactly. Bad JSON fails fast at import. - docs/templates/guided_review.md — the Phase 5 acceptance worksheet: per-question answer/citation verdicts against a rubric, the metadata-filter check, a knob-change log, and a decision block that becomes the client's validation summary. - CLIENT_DEPLOYMENT_RUNBOOK.md Phase 5 tightened into a lightweight-path checklist pointing at the worksheet; the two new env vars documented in .env.example and the Environment reference. PIPELINE_TUNING.md / ARCHITECTURE.md / CLAUDE.md updated to name the config instead of "edit the source". evaluation/client/ README.md lists the worksheet as the default artifact. Tests: tests/test_pipeline_constants.py (_json_env: default / blank / override / bad-JSON / behaviour-preserving defaults); per-client vocab override cases in test_query_pipeline.py. Closes #135 Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016Pcofh3BK8CAoHXcfEZoQj
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Steps completed
pipeline/constants.py—METADATA_FILTER_DOCTYPE_KEYWORDS/METADATA_FILTER_STATUS_KEYWORDSas JSON env vars via_json_env()(fails fast on bad JSON). Status form is a{keyword: status-value}map, so a client can map"expired"→"Inactive". Defaults reproduce the previous hardcoded behaviour exactly.pipeline/query_pipeline.py—_extract_filters()reads the config; the_DOCTYPE_KEYWORDS/_STATUS_KEYWORDSmodule constants are gonedocs/templates/guided_review.md— the Phase 5 worksheet: rubric, per-question answer/citation verdicts, metadata-filter check, knob-change log, decision blockdocs/CLIENT_DEPLOYMENT_RUNBOOK.mdPhase 5 → lightweight-path checklist pointing at the worksheet; two new env vars in the Environment reference +.env.exampledocs/PIPELINE_TUNING.md§ Per-client,docs/ARCHITECTURE.md,CLAUDE.md— name the config instead of "edit the source";evaluation/client/README.mdlists the worksheet as the default artifactCloses / Refs
Closes #135
Refs #127
Test plan
pytest tests/— 223 passed / 67 skipped, no networkruff check .passestests/test_pipeline_constants.py—_json_envdefault / blank / override / bad-JSON / behaviour-preserving-defaultstests/test_query_pipeline.py— per-client doctype + status vocab overrides viamonkeypatch🤖 Generated with Claude Code