slide-reviewer
Skill0 starsChanged 12 days agoMITView on GitHub
Critique web-slides deck screenshots after npm run review — overlap, density, weak titles, safe margins, contrast feel — and write findings to review/critique.md. Does not auto-rewrite slides unless the user asks. Use after screenshots exist and before PDF export, or when the user asks for a visual review / design critique of the deck.
In this repository2 other files it gives its agents.
- style-librarySkill
- web-slidesSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
When the user asks “does this look good?”, “critique the deck”, or “reviewer pass”.
- Very specific: line 38
Ensure review/report.md and review/slide-*.png (and contact-sheet.png) exist.
- Very specific: line 39
Run npm run critique (or the script above).
- Very specific: line 42
Open the contact sheet and every slide PNG: Use the rubric in references/critique-rubric.md.
- Very specific: line 45
If they ask to fix: address blockers first, re-run npm run review + npm run critique, repeat until they accept.
- Distinctive: line 72
No automatic TSX edits.
- Distinctive: line 73
No PDF export (that stays web-slides step after critique acceptance).
- Distinctive: line 74
No inventing facts or changing sourced figures during critique.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Slide reviewer
A critique phase, not a silent fixer. After npm run review produces screenshots, study them (and the lint report), write structured findings, and stop. Only rewrite slides when the user asks you to address specific findings.
When to use
- Immediately after web-slides workflow Screenshots and review, before Export.
- When the user asks “does this look good?”, “critique the deck”, or “reviewer pass”.
Commands
From a deck directory (script ships in the template):
npm run review # screenshots + lint → review/
npm run critique # aggregate lint + scaffold critique.md
# then the agent opens PNGs and fills visual findingsFrom this repo without a deck script path:
node skills/slide-reviewer/scripts/critique.mjs --deck <deck-dir>Agent workflow
- Ensure
review/report.mdandreview/slide-*.png(andcontact-sheet.png) exist. If not, runnpm run reviewfirst. - Run
npm run critique(or the script above). It writes / refreshesreview/critique.mdwith:- Automated findings from
report.md(overlap, safe-margin, tiny-text, contrast, title-sentence, …) - Empty Visual findings sections for you to complete
- Automated findings from
- Open the contact sheet and every slide PNG. Use the rubric in
references/critique-rubric.md. - Append or edit findings under Visual. Each finding:
severity(blocker / warn / note),slideid or number,issue,evidence(what you saw),suggestion(optional — do not implement yet). - Show the critique to the user. Do not auto-rewrite TSX/YAML unless they ask to fix items (by severity or by id).
- If they ask to fix: address blockers first, re-run
npm run review+npm run critique, repeat until they accept.
What to critique
| Area | Look for |
| Overlap | Text through cards, peer boxes colliding, footer collisions |
| Density | >~40 words, cramped KPI rows, no whitespace |
| Titles | Not a full-sentence takeaway; weak verbs; section labels as titles |
| Safe margins | Content in padding band; edge-hugging type |
| Title hero | Centered headline over busy photo; missing left scrim; low contrast on scrim |
| Hierarchy | Competing focals; accent used everywhere; more than one hero effect per slide |
| Default look | Untouched template theme and tokens the user never chose |
| Spec drift | Approved deck.spec.yaml types/order vs what screenshots show |
Automated lint already covers many of these as errors/warnings in report.md — the critique aggregates them and adds judgment the lint cannot make.
Outputs
| File | Role |
review/report.md | Machine lint from npm run review |
review/critique.md | Human+agent critique (this skill) |
review/contact-sheet.png | First look |
Non-goals
- No automatic TSX edits.
- No PDF export (that stays web-slides step after critique acceptance).
- No inventing facts or changing sourced figures during critique.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

