Repository navigation
Chunk Quality v3.2: label overlap, add Chunks-view legend - #10
Merged
Merged
Conversation
Chuck couldn't tell intentional RAG overlap from duplicate content, and
the Chunks view assumed vocabulary it never defined. All deterministic —
no AI calls.
API (api/chunk.js):
- addOverlap() records overlap_word_count on each section it modifies;
enhanceChunks() carries it into the response
- settings.target_words exposes the active {min,max,target} size range
so the UI reads it live instead of mirroring CHUNK_SIZES
UI (public/index.html):
- Overlap prefix renders dimmed with a "↔ Nw overlap" badge + tooltip
explaining the intentional carry-over
- Collapsible "How to read this" legend defines Chunk, L#, Path,
Sections, w/t, live target size range, and overlap
- Chunk meta pill gets an explanatory tooltip; "1 sections" plural fixed
- Heading-path breadcrumb labeled "PATH" with tooltip + legend entry
- Content Summary stat labels swapped (big=Chunks, small=Sections) to
match the chunk cards and legend
Smoke-tested on the page Chuck screenshotted (Hayes Barton Place
assisted-living: Chunk 2's 29-word overlap now renders labeled) plus
searchinfluence.com, Wikipedia RAG article, and getchunks itself.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
type="url" made the browser reject bare domains before submit. Switch to type="text" (inputmode="url" keeps the mobile keyboard) and default to https:// in the submit handler when no scheme is given. Full URLs pass through untouched; the API's scheme allowlist stays strict. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The title attribute gave a help cursor but no visible tip in practice — native tooltips need a long hover rest and never render in some embedded browsers. New [data-tip] component renders the hint as a styled ::after immediately on hover (.tip-right modifier anchors the chunk meta pill's tip to its right edge). big-chunk-meta drops overflow:hidden/ellipsis so the pseudo-element isn't clipped; it wraps instead when narrow. The overlap prefix keeps its native title as a best-effort extra — CSS tooltips don't position reliably on multi-line inline spans, and the badge beside it carries the same message. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This was referenced Jul 25, 2026
This branch was successfully deployed
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
1 participant
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.
Why
Chuck's feedback: the Chunks view was unreadable to a non-dev — the RAG overlap prefix looked like duplicate content, and the vocabulary (Chunk, L#, sections, w/t, the breadcrumb) was never defined. Phase 1 of the Chunk Quality plan (
tasks/todo.md) makes the current output self-explanatory. Everything is deterministic — zero AI/LLM calls.API (
api/chunk.js)addOverlap()now recordsoverlap_word_counton each section it modifies, carried throughenhanceChunks()into the responsesettings.target_wordsexposes the active{min, max, target}word range so the UI reads size ranges live instead of duplicatingCHUNK_SIZESUI (
public/index.html)↔ Nw overlapbadge; tooltip explains it's intentional RAG-style context carry-over, not duplicate contentChunk 1 • L2 • 2 sections • 105w • 115t) gets a tooltip spelling out each part; fixed the1 sectionspluralTesting
Smoke-tested locally (
vercel dev) with Playwright at 1440px:overlap_word_count: 29and renders the 29-word prefix badged + dimmed; 7 badges total (2w–35w)target_wordstracks the auto-detected size (small/medium/large), zero-overlap pages show no badgesNext: Phase 2 (v3.3) adds the actionable layer — deterministic per-chunk flags + a 0–100 Chunkability score.
🤖 Generated with Claude Code