feat(blog): improve cover image generation and refresh blog artwork - #8156
Open
ankur-arch wants to merge 9 commits into
Open
feat(blog): improve cover image generation and refresh blog artwork#8156ankur-arch wants to merge 9 commits into
ankur-arch wants to merge 9 commits into
Conversation
Message-first workflow (extract primary message + core highlight before any drawing), derived cover titles separate from article titles, component library (chart/metric/table/code/query/flow/object/prism/field/type) with article-to-visual routing, gradient/light system with SVG recipes on the eclipse spectrum tokens, modular template families, batch distribution + contact-sheet review, and rebrand-corrected design-review criteria. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…rection system Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Each cover derives from its article's core message and strongest visual highlight, routed to a component family with one light treatment. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
embed-fonts.py collected glyphs with a regex that only matched a text run immediately followed by a closing tag, so any run preceding a nested tspan fell out of the subset and Chrome rendered it in a fallback serif - silent, visible only in the raster. Flatten tspans before collecting, and decode entities. Twelve affected covers re-embedded and re-exported. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Contributor
|
Important Review skippedToo many files! This PR contains 246 files, which is 96 over the limit of 150. To get a review, reduce the PR to 150 files or fewer by splitting it into smaller PRs or changing its base branch. Upgrade to Pro+ to raise the limit. This review couldn't start because sufficient usage credits or metered capacity aren't available. Add credits or update usage-based reviews in the billing tab, then retry. ⚙️ Run configurationConfiguration used: Repository UI Review profile: CHILL Plan: Pro Run ID: ⛔ Files ignored due to path filters (481)
📒 Files selected for processing (246)
You can disable this status message by setting the |
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Whole-archive contact-sheet review found the prism-refraction motif used on ten covers and four duplicated headlines. Redesigned thirteen covers into distinct component families so cohesion comes from type, color, surface and restraint rather than a shared composition. Co-Authored-By: Claude Opus 5 (1M context) <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
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.
What changed
The cover-image skill is now an editorial art-direction system rather than a template filler, and every blog cover in the archive has been regenerated through it.
Skill upgrades (
.claude/skills/content-create-hero-image/)references/component-library.md): HeroChart, MetricComponent, ComparisonTable, CodeWindow, QueryPanel, ArchitectureFlow, DatabaseObject, PrismObject, GradientField, TypeMark, with an article→visual routing heuristic and calibration against the strongest developer-brand OG art.references/gradient-system.md): spectral wash, radial bloom, product glow, prism/spectrum gradient, spectral edge, gradient beam, mesh field, aurora band, fade mask — each an SVG recipe on real eclipse tokens, with the "what is this light doing?" test replacing a blanket no-gradients rule.assets/templates/): editorial, chart, code, table, flow, object, abstract. They share canvas, typography, tokens, primitive gradient defs, and the inlined lockup — not one composition.references/design-review.md): rebrand-corrected type and color categories plus new criteria for message fidelity, highlight discipline, gradient quality, accuracy, and editorial quality; every cover is reviewed at full size and at ~300px thumbnail size.scripts/validate-covers.sh(dimensions, size budget, embedded fonts, frontmatter wiring) andscripts/contact-sheet.sh(thumbnail contact sheets for whole-archive review).Design approach
Each cover is now driven by the article's core message and strongest visual highlight rather than by a fixed layout. The routing heuristic is a starting point; the article's actual message overrides it. Composition follows one idea, one hero, one highlight, plenty of space — one dominant component, one light treatment, deliberate negative space, and a viewer who can find the point in under a second. Consistency across the archive comes from typography, prism color, paper surfaces, spacing, and restraint — deliberately not from a shared layout.
Competitive calibration informed the bar: the strongest developer-brand covers carry one or two elements, use an object bespoke to the article's idea, and never screenshot product UI. Dark is the crowd, so Prisma's light paper surface is treated as the differentiator.
Validation
apps/blog/content/blognow carry a current-brand cover.embed-fonts.py→ headless-Chromeexport-png.sh), opened as a raster, graded against the design-review categories, fixed worst-first, and re-rendered. The raster is the product; nothing shipped on "the SVG compiles".validate-covers.sh apps/blog --allpasses for all 296 posts: exact 1200×630 dimensions, embedded subset fonts, meta PNG under the 1 MB budget, and frontmatter wired tohero.svg/meta.png.pnpm buildinapps/blogcompiles and prerenders all 671 pages (the search route needsMIXEDBREAD_API_KEY, a pre-existing requirement). Zero broken image references across all frontmatter paths.Component families used
Charts, single-metric compositions, comparison tables, code windows, query/trace panels, architecture flows, database objects, prism objects, gradient fields, and typography-led covers — plus bespoke one-off graphics where an article's idea deserved its own form (a freshness window, a scan-scope diagram, a hash-bucket, an origin boundary, a file tree).
Notes and limitations
heroImagePath,metaImagePath, andheroImageAlt. No titles, slugs, dates, or body content were modified.Dependency
This PR is stacked on top of Shane's PR: #8142
Its base branch is
shane/launch. It contains no duplicate of Shane's changes.