improve-animations
An agent skill by emilkowalski, from emilkowalski/skills. Tags: analysis, animation, code-quality, planning, review.
What it does
Survey a codebase's animation and motion code as a senior motion advisor, then produce a prioritized audit and self-contained implementation plans for other agents (or cheaper models) to execute. Read-only on source code — it plans improvements, it does not apply them. Use when the user asks to "improve the animations", "audit the motion", "make this app feel better", or wants a roadmap of animation fixes rather than a review of a single diff.
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add emilkowalski/skills --skill improve-animations
Or copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):
git clone --depth 1 https://github.com/emilkowalski/skills
cp -r skills/skills/improve-animations ~/.claude/skills/improve-animations
Safety box score
Grade B, low risk: 12 of 100 risk points, rated 2026-10-20. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.
This skill is a read-only animation auditor and planner for codebases, with a separate `execute` mode that can hand implementation to another agent in an isolated worktree. It does not touch machine-wide settings or secrets, but it does read arbitrary repo content and can write plan documents under `plans/`, so the main risks are prompt-injection exposure and limited repo mutation.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 0 (none) | The shipped files are plain Markdown and contain no native scripts, `eval`, or shell execution logic; the only command shown is a documentation example (`git rev-parse --short HEAD`) in a template. |
| Network access | 0 (none) | There is no code that downloads from or sends data to remote hosts; the repo content is read locally and the links are informational only. |
| Filesystem reach | 1 (low) | The skill is read-only for source, but it explicitly allows creating and editing plan files under `plans/` and `plans/README.md`, which is a bounded write capability inside the repo. |
| Secrets & credentials | 0 (none) | It does not request or access tokens, passwords, browser sessions, `.env` files, or key stores. |
| Privilege & persistence | 0 (none) | No sudo, global installs, services, shell startup edits, or OS-level configuration changes are described. |
| Instruction hijack surface | 1 (low) | It explicitly reads arbitrary repository content and calls out hostile text like `ignore previous instructions…` as something to flag, which means the audit surface includes prompt-injection attempts in repo files. |
| Irreversible actions | 1 (low) | The `execute` variant can dispatch another agent to implement a plan in an isolated worktree, so it can make repo changes, but it does not push, deploy, or perform external actions. |
| Transparency | 0 (none) | All shipped files are readable Markdown; there are no binaries, packed payloads, or code decoded at runtime. |
Before installing
- Verify you are okay with the skill writing only under `plans/` and `plans/README.md`, not source files.
- If you want strict read-only behavior, avoid the `execute` mode and use audit/plan-only workflows.
- Test it once in a scratch repo that contains hostile text like `ignore previous instructions` to confirm the prompt-injection handling is acceptable.
Rated 2026-10-20 by an automated review of 3 files from the skill's repository.
Source
- Repository
- emilkowalski/skills (all skills from this repository)
- Path
- skills/improve-animations/SKILL.md
- Branch
- main
- Updated
- 2026-09-22
Related skills
- differential-review — Performs security-focused differential review of code changes.
- principle-minimize-reader-load — Apply when reviewing or shaping code that's hard to trace.
- reflect — Spawn three parallel review subagents over the active transcript, surface learnings, and route each to a concrete edit on an existing skill.
- run-learning-retrospective — Evaluate learning progress, identify blockers, and adjust the learning plan
- thermo-nuclear-code-quality-review — Run an extremely strict maintainability review for abstraction quality, giant files, and spaghetti-condition growth.
- weekly-review — Produce a weekly synthesis of authored commits with highlights by bugfix, tech debt, and net-new work