docs: sync docs/DESIGN.md's Pending/Open-questions with Milestone 1's actual state - #14
Merged
Merged
Conversation
Both still described Milestone 1 as in-progress ("just needs a live
run", "open now: finish the spot-check") even though D16 in the same
file documents it as fully run (20/20 titles), hand-read, and closed
with two confirmed leak mechanisms and judge iteration ended.
Co-Authored-By: Claude <noreply@anthropic.com>
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Scheduled documentation-sync pass. Compared README.md and docs/DESIGN.md against the actual working tree (src/preshow/, evals/, webapp/, content/researched/, titles.yaml) rather than trusting CLAUDE.md's own Status prose or commit history.
Result
README.md was already accurate — Milestone 1 completion, the closed six-judge search, the 8/20 researched vs. 20/20-TMDB split, env vars, project layout, and the "26 tests" claim all matched the code (
grep -c "^def test_" tests/= 26). No changes needed there.docs/DESIGN.mdhad one genuine internal inconsistency: its own Pending and Open questions sections still described Milestone 1 as in-progress ("just needs a live run", "open now: finish the human spot-check... then decide whetherleakage_rate... is trustworthy enough to report") — even though D16, in the same file, documents Milestone 1 as fully run (20/20 titles), hand-read, closed with two confirmed leak mechanisms, and judge iteration ended for good (a sixth judge,SimilarityJudge, also failed live). Updated both sections to match D16's actual conclusion.Verified directly against the working tree
src/preshow/retrieval.pymatches the described GREEN/AMBER/RED mechanism exactly (overview/production/accolades GREEN, plot never constructed as a source, reception treated as RED).Substring,LLM,NLI,TrainedClassifier,Hybrid,Similarity) exist inevals/judge.py.content/researched/has exactly 8 titles;evals/dataset/titles.yamlhas 20 (matches the README's 8-researched/20-total claim).UPSTASH_REDIS_REST_URL/TOKEN) are read but optional insrc/preshow/kv_store.py, falling back to local files — matches D11/README.Not re-run: pytest isn't installed in this environment, and this project's own CLAUDE.md documents a known, recurring Groq network-access issue in Claude Code Bash sessions — no eval/test numbers in this PR are freshly reproduced, only cross-checked against code structure.
No code defects found during this pass.
Generated by Claude Code