Repository navigation
Conversation
Codex Review Summary: ❌ FailedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
|
Example report and chart gallery on Drop, both built with |
133ee1f to
4ef8921
Compare
comark-annotate
comark-badge
comark-cards
comark-carousel
comark-cast
comark-charts
comark-chat
comark-chips
comark-cjk
comark-code
comark-code-notes
comark-colors
comark-comments
comark-critic
comark-deflist
comark-details
comark-diagrams
comark-diff
comark-embed
comark-fields
comark-figure
comark-file-tree
comark-footnotes
comark-github
comark-icon
comark-include
comark-keys
comark-links
comark-linter
comark-magic-move
comark-mark
markdown-it-to-comark
marked-to-comark
comark-mathml
comark-media
comark-metrics
migrate-to-comark
comark-outline
comark-properties
comark-qr
comark-quiz
comark-reading-time
remark-to-comark
comark-sources
comark-steps
comark-table
comark-tabs
comark-tests
comark-time
comark-ui
comark-xref
comark-zoom
commit: |
![pullfrog[bot]](https://avatars.githubusercontent.com/in/1768019?s=60&v=4){kind=small}
There was a problem hiding this comment.
Important
Section tracking does not work with the default renderer, and the documented build setup needs two corrections to run successfully from a clean environment.
Reviewed changes — Reviewed the new data-report skill, its standalone HTML builder, references, and documentation integration.
- Dataset-to-report workflow: Adds instructions for profiling data, computing reproducible figures, choosing charts, and delivering the report with its analysis script.
- HTML report builder: Adds sticky section navigation, an outline and metadata sidebar, light/dark switching, issue rejection, and optional screenshots.
- Examples and integration: Adds the chart gallery, support-analytics example, skill metadata, docs page and index entry, and editor round-trip coverage.
Validation: workspace build and docs lint passed; focused tests passed with 586 tests and one skipped. Built both references, checked desktop/mobile rendering and theme switching, and exercised screenshot generation and invalid-input rejection.
gpt-6.1-sol(xhigh) ｜ 𝕏
Latest result Pull request closed. Final answersSession Pull request closed. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. Session Waiting for open parent PR #406 to merge; this child has nothing left to repair. |
cb63d5c to
5548820
Compare
a429b26 to
79b5c8a
Compare
There was a problem hiding this comment.
✅ No new issues found. No P0 or P1 findings remain.
Reviewed changes — Reviewed the changes since the prior Pullfrog review, including the navigation repairs, build instructions, refreshed example, and newly added recipe.
- Fixed section navigation: Resolved outline targets by ID and applied scroll offsets to wrapped headings, restoring current-section tracking and positioning headings below the sticky header.
- Corrected build setup: Documented copying the builder beside
report.mdand installing Chromium before generating screenshots. - Added navigation regression tests: Covered the generated report's seven outline targets, active links in both navigation areas, and fragment positioning.
- Added the report recipe: Registered the complete report as a playable preset and refreshed the example's figures and answerability chart.
Validation: both navigation E2E tests passed, including repeated parallel runs; focused editor, skill, and workspace tests passed with 591 tests and one skipped. Docs lint and focused E2E lint passed. The recipe loaded and rendered eight charts, the copied builder produced both screenshots, and desktop/mobile fragment navigation left headings below the header. Full workspace build attempts reached [timed out after 120000ms]; direct UI and linter builds passed.
gpt-6.1-sol(xhigh) ｜ 𝕏
79b5c8a to
d37839d
Compare
…s page with a top bar and a sidebar
The build names a missing package instead of throwing, and saves screenshots in slices with the Data folds closed. The skill says how to handle several sources, prints the stats table from the script, counts series and bars apart, says to install a current comark-ui when props are refused, and adds a spread-per-name chart.
d37839d to
7d91804
Compare
There was a problem hiding this comment.
ℹ️ No P0 or P1 findings remain. One small dependency-diagnostic fix is noted inline.
Reviewed changes — Reviewed the incremental changes since the prior Pullfrog review, excluding unchanged content carried through the rebase.
- Refined the analysis workflow: Clarified multiple-source profiling, generated headline tables, category limits, metadata, and screenshot checks.
- Expanded chart guidance: Added the median/p90 dumbbell example and clarified palette ordering, stacked labels, supported props, and linear money comparisons.
- Extended dependency resolution: Added report-directory fallback and installation guidance when a required module cannot be resolved.
- Sliced screenshot output: Saved numbered light/dark images up to 1400px tall while leaving Data folds closed.
Validation: workspace build and docs lint passed; focused editor, skills, and workspace tests passed with 591 tests and one skipped; both navigation E2Es passed. The copied builder resolved dependencies from the report directory and generated both references with correctly sized slices in both themes. The missing-preset case reproduced the diagnostic defect below.
gpt-6.1-sol(xhigh) ｜ 𝕏
| } catch {} | ||
| } | ||
| console.error( | ||
| `Can't find ${name} from this folder or the report's. Install it here: pnpm add ${name}`, |
There was a problem hiding this comment.
P2 · Class: dependency diagnostics treat import subpaths as package names. With comark and @comark/html installed but the preset missing, this prints pnpm add comark-ui/html; pnpm interprets that as the GitHub repository comark-ui/html, not the npm package, so following the recovery instruction does not install the dependency. Suggest pnpm add comark-ui for this import while preserving @comark/html as a scoped package name.
Someone with a large export and no BI tool has no path in comark-ui to the kind of page dashboards such as nuxt-evals and Slopalytics show: a strip of headline figures, a few findings, a section of charts per question, and a sidebar. The
reportskill writes a decision page from research, so it starts after the analysis, not from raw data.This adds a
data-reportskill. It walks an agent from raw data to the page: profile the data, write four to seven questions, compute every number in one script it keeps, pick each chart from a house style, then build and screenshot the page.scripts/build.mjswraps the Markdown in one self-contained HTML file. The file has a sticky top bar with a link per section, the report, and a sidebar with the outline and the frontmatter's facts about the data. Light and dark follow the system, with a toggle, and the sections in view are marked as you scroll. The build refuses a report with issues and prints each one.references/charts.mdmaps each question shape to its chart and is a valid report itself, so the same script builds it into a chart gallery.references/example.mdis a finished support-analytics report over made-up data, built from a seeded script so its numbers agree across charts. The skill gets a docs page, a line on the skills index, and a place in the editor test over every skill's example. A new "A data report" recipe holds the whole example, so the playground opens it as a preset next to the charts'.This needs #406's chart props (
heatmap,unit,log,description), so it's based on that branch.I ran the build as the skill tells an agent to, in a fresh folder with comark-ui linked from this branch. It wrote the page and both screenshots, and refused a report with
better="more". The ui build,editor.test.ts,workspace.test.tsand the docs lint pass. I also gave a fresh agent only this skill and the public export of this repository's 400 pull requests and last 1,000 Actions runs. It built the report in about 7 minutes and listed every place it had to guess; the fixes are in the last commit, and the rebuilt report is at https://drop.vitehub.dev/d/wrhphb6sqz1w/data-report-skill-test-comark-ui-github. With comark-ui from npm the build still refuses the page, sincedescription,unitandlogship with #406. I haven't checked the page on a phone; under 1100px the sidebar moves under the report and under 640px the top bar drops its links.