Simplify
π€ Curated agent skills for AI agents β one coherent engineering loop that compounds: grill β spec β implement β review β commit β learn. Adopt it or fork it!
npx -y skills add toverux/grimoire --skill simplifyAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 27 days oldThe repository was created 27 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Optional pre-review quality pass β behavior-preserving fixes for reuse, dead weight, altitude, and clarity. Bug hunting is /review-gate's job.
SKILL.md
4.2 KB, as published. Nobody here has run it
Simplify recently changed code while preserving exact behavior: same output for every input, same error behavior, same side effects and ordering. Prioritize readable, explicit code over compact code β fewer lines is not the goal.
Step 1: Resolve the scope
- If the user named a scope (a file, a directory, "the function I just wrote"), use it as authoritative β do not widen it.
- Otherwise, use the diff between the current branch and its merge-base with the default branch (
git diff <default-branch>...). With no base ref, fall back to staged + unstaged changes (git diff HEAD). - Outside a git repository, use the files edited earlier in this conversation.
If the scope comes up empty, stop and ask the user what to simplify.
Preflight: the reviewers hunt in code. If the scope is documentation-only, or only generated, vendored, lockfile, or mechanical churn (formatting, lint autofix, mass rename), stop with a one-line note that there is nothing to simplify. On a mixed diff, narrow the scope to the code files and continue. Gate on the kind of change only β a user-named scope always runs.
Step 2: Launch three reviewers in parallel
Dispatch one subagent per persona, in parallel where the harness supports it, sequentially otherwise. For each, read its file and pass the full file content verbatim as the subagent's prompt, together with the resolved scope (the full diff or file set) β paraphrasing from memory loses the gating rules that keep the pass behavior-preserving:
references/personas/code-reuse-reviewer.mdβ existing utilities, duplicated functionality, reimplemented stdlib primitives.references/personas/code-quality-reviewer.mdβ dead weight, altitude (abstraction level), clarity.references/personas/efficiency-reviewer.mdβ wasted work, missed concurrency, memory.
Step 3: Apply fixes
Wait for all three reviewers, aggregate their findings, and fix each issue directly. A false positive or a fix not worth its churn: note it, skip it, move on β settle it yourself rather than raising it to the user.
Before applying each fix, confirm it preserves behavior (the test above). If it can't clear that test, skip it.
Never simplify away a safety check. Input validation at trust boundaries, error handling that prevents data loss, security checks (authorization, escaping, sanitization), and accessibility affordances stay β even when a finding frames them as redundant. Code that drops one is not simpler, it is unfinished.
Step 4: Verify behavior is preserved
- Typecheck and lint the full project β they catch the common simplification regressions: broken imports, dropped narrowings, dead code other modules still reference.
- Run tests scoped to the blast radius of the changes; broaden when shared or heavily-imported code was touched. No scoping mechanism β full suite.
On a failure, fix the underlying break or revert the specific simplification that caused it β weakening assertions, types, or skipping tests defeats the behavior-preservation guarantee. If no checks are configured, state that in the summary.
Step 5: Summarize
Report fixes applied per dimension (reuse, quality, efficiency), findings skipped as false positives or not worthwhile, and which checks ran with their results. The measure is what improved and that behavior held β many clarity and safety fixes preserve or add lines.
Close with a flow pointer (presentation): /review-gate (user-invoked) β the gate that hunts for bugs and spec drift, in this session; suggest low for a trivial or mechanical diff, high for a large, cross-cutting, or risky one, medium otherwise.
Gives 0 of the 12 instructions most refactoring skills give
Counted across 521 of the 525 authors here whose files we hold, read 2026-08-06
- run tests after each changein 59 of 521, across 56 files
- write tests before refactoringin 27 of 521, across 24 files
- preserve external behaviorin 26 of 521, across 22 files
- remove dead codein 25 of 521, across 24 files
- make small incremental changesin 20 of 521, across 17 files
- break the implementation into tiny commitsin 18 of 521, across 5 files
- ask the user about alternative optionsin 17 of 521, across 4 files
- create a GitHub issue with the planin 17 of 521, across 4 files
- explore the repository to verify assertionsin 17 of 521, across 4 files
- interview the user about the refactorin 16 of 521, across 3 files
- check the codebase for test coveragein 16 of 521, across 3 files
- refactor one thing at a timein 16 of 521, across 12 files
Said here and by no other author read
- skip non-code files during preflight
- pass the full file content verbatim to reviewers
- typecheck and lint the full project
- run tests scoped to the blast radius
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.