Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Plan, scaffold, and render Manim Community Edition explainer videos for math, algorithms, technical concepts, and data stories.
.claude/skills/cowork-os-manim-video/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-01 | ✓→✗ | ▼ Worse | -80% | 0% |
| case-03 | ✓→✗ | ▼ Worse | -81% | 0% |
Build production-ready Manim CE explainer videos with a stronger workflow than a plain prompt: dependency preflight, deterministic project scaffolding, render helpers, and artifact manifests.
plan.md, script.py, render helpers, and run artifacts.| Name | Type | Required | Description | |---|---|---|---| | topic | string | Yes | What the animation should explain or visualize | | mode | select | No | Video pattern: auto, concept-explainer, equation-derivation, algorithm-visualization, data-story, architecture-diagram, paper-explainer, 3d-visualization | | audience | string | No | Target audience for pacing and explanation depth | | target_length_seconds | string | No | Approximate runtime in seconds | | output_dir | string | No | Workspace-relative or absolute project directory | | voiceover | select | No | Whether to scaffold a voiceover script |
scripts/setup.sh first. This is mandatory before claiming the project is render-ready.scripts/bootstrap_project.py to create or refresh the initial project shape instead of hand-rolling boilerplate.plan.md and script.py aligned. If the scene list changes, update both.references/full-guidance.md: planning rubric, scene design rules, code patterns, render flow, and review checklistreferences/troubleshooting.md: install issues, LaTeX/render failures, and debug movesscripts/setup.sh: environment preflight for Python, Manim, LaTeX, and ffmpegscripts/bootstrap_project.py: deterministic scaffold generator for a new Manim project../manim-video.json.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→fail | 25,084 | 5,633 | -78% | 1 | 1 | 0% | 4,869 | 978 | -80% | 0 | 0 | — |
case-02 | pass→pass | 28,413 | 27,038 | -5% | 1 | 1 | 0% | 5,788 | 6,888 | +19% | 0 | 0 | — |
case-03 | pass→fail | 27,962 | 6,097 | -78% | 1 | 1 | 0% | 5,463 | 1,057 | -81% | 0 | 0 | — |
case-04 | pass→pass | 5,453 | 8,902 | +63% | 1 | 1 | 0% | 928 | 2,101 | +126% | 0 | 0 | — |
case-05 | pass→fail | 20,016 | 3,852 | -81% | 1 | 1 | 0% | 3,695 | 901 | -76% | 0 | 0 | — |
case-06 | pass→pass | 11,787 | 13,227 | +12% | 1 | 1 | 0% | 1,957 | 2,860 | +46% | 0 | 0 | — |
case-07 | pass→pass | 9,063 | 5,869 | -35% | 1 | 1 | 0% | 1,583 | 1,633 | +3% | 0 | 0 | — |
case-08 | pass→pass | 19,465 | 21,205 | +9% | 1 | 1 | 0% | 3,332 | 4,198 | +26% | 0 | 0 | — |
case-09 | pass→pass | 15,801 | 10,303 | -35% | 1 | 1 | 0% | 2,554 | 2,181 | -15% | 0 | 0 | — |
case-10 | pass→pass | 11,588 | 7,649 | -34% | 1 | 1 | 0% | 1,970 | 1,932 | -2% | 0 | 0 | — |
case-11 | pass→fail | 13,839 | 5,644 | -59% | 1 | 1 | 0% | 2,581 | 962 | -63% | 0 | 0 | — |
case-12 | fail→fail | 6,183 | 7,510 | +21% | 1 | 1 | 0% | 273 | 1,025 | +275% | 0 | 0 | — |
case-13 | fail→fail | 29,294 | 7,015 | -76% | 1 | 1 | 0% | 5,684 | 1,017 | -82% | 0 | 0 | — |
case-14 | pass→pass | 22,878 | 27,939 | +22% | 1 | 1 | 0% | 3,681 | 4,943 | +34% | 0 | 0 | — |
case-15 | fail→fail | 32,102 | 30,971 | -4% | 1 | 1 | 0% | 6,161 | 6,846 | +11% | 0 | 0 | — |
case-16 | pass→fail | 19,428 | 6,688 | -66% | 1 | 1 | 0% | 3,738 | 904 | -76% | 0 | 0 | — |
case-17 | fail→pass | 31,418 | 26,581 | -15% | 1 | 1 | 0% | 6,153 | 5,648 | -8% | 0 | 0 | — |
case-18 | fail→fail | 24,990 | 6,381 | -74% | 1 | 1 | 0% | 4,738 | 1,058 | -78% | 0 | 0 | — |
case-19 | fail→pass | 30,602 | 30,028 | -2% | 1 | 1 | 0% | 6,151 | 5,963 | -3% | 0 | 0 | — |
case-20 | fail→pass | 13,919 | 2,897 | -79% | 1 | 1 | 0% | 2,270 | 1,063 | -53% | 0 | 0 | — |
case-21 | pass→pass | 15,463 | 8,252 | -47% | 1 | 1 | 0% | 2,638 | 2,119 | -20% | 0 | 0 | — |
case-22 | fail→fail | 28,828 | 16,257 | -44% | 1 | 1 | 0% | 6,173 | 4,130 | -33% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 15 counted toward the lift figure. The other 7 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of -9 percentage points is the difference between those two pass rates over the 15 comparable cases. 6 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.