distill
Skill120 starsChanged todayMITView on GitHub
Cut the branch down to what the code cannot say: reuse what exists, remove duplication, dead code, and wasted work, turn invariants into asserts, and delete the comments the code already speaks. Run it after answering review feedback too — a fix written to satisfy a reviewer is the least distilled code on the branch.
76 other projects use a version of this.
Adopted by
Most recently taken up by
- DeepExtrema/Tesseracttheir own version, first seen Nov 16, 2026
- persiyanov/skillstheir own version, first seen Nov 15, 2026
- komluk/scaffoldingtheir own version, first seen Nov 15, 2026
- Integralist/agent-skillstheir own version, first seen Nov 14, 2026
- theexperiencecompany/gaiatheir own version, first seen Nov 13, 2026
- venhdev/skillstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use layout, shape and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
owshq-mecSkilllayout
Improve layout, spacing, and visual rhythm. Fixes monotonous grids, inconsistent spacing, and weak visual hierarchy. Use when the user mentions layout feeling off, spacing issues, visual hierarchy, crowded UI, alignment problems, or wanting better composition.
Also in 8 projects that use this
HAR5HA-7663Skillshape
Plan the UX and UI for a feature before writing code. Runs a structured discovery interview, then produces a design brief that guides implementation. Use during the planning phase to establish design direction, constraints, and strategy before any code is written.
Also in 11 projects that use this
ReScienceLabSkillarchive
Archive session learnings, debugging solutions, and deployment logs to .archive/yyyy-mm-dd/ as indexed markdown with searchable tags. Use when completing a significant task, resolving a tricky bug, deploying, or when the user says \"archive this\". Maintains .archive/MEMORY.md index for cross-session knowledge reuse.
Also in 3 projects that use this
klauswgSkillckm:design-system
Token architecture, component specifications, and slide generation. Three-layer tokens (primitive→semantic→component), CSS variables, spacing/typography scales, component specs, strategic slide creation. Use for design tokens, systematic design, brand-compliant presentations.
Also in 3 projects that use this
iovisorSkillevolve
Evolve uBPF through a gated spec-first workflow that propagates approved changes through requirements, design, validation, implementation, and RFC alignment artifacts.
Also in 3 projects that use this
termdockSkillrecall
Retrieves cross-session memories about past decisions, lessons, and patterns. Use when user asks about prior work, past decisions, or mentions a feature. Triggers include check memory, what did we decide, how did we solve.
Also in 5 projects that use this
Other versions76 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 76
- juxt/allium★ 506same name
- educlopez/ui-craft★ 441same name
- Michaelliv/napkin★ 428same name
- nexmoe/DayBee★ 424same name
- nexmoe/eve★ 424same name
- Southclaws/storyden★ 351same name
- 10xChengTu/Mux0★ 337same name
- theexperiencecompany/gaia★ 327same name
- AkaraChen/aghub★ 271same name
- zacharyfmarion/openscad-studio★ 252same name
66 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository39 other files it gives its agents.
- AGENTS.mdAGENTS.md
- benchmark/AGENTS.mdAGENTS.md
- docs/AGENTS.mdAGENTS.md
- package-gale/AGENTS.mdAGENTS.md
- package-grog/AGENTS.mdAGENTS.md
- package-jade/AGENTS.mdAGENTS.md
- package-loam/AGENTS.mdAGENTS.md
- package-re/AGENTS.mdAGENTS.md
- package-wadopoet/AGENTS.mdAGENTS.md
- wado-bundled-icu/AGENTS.mdAGENTS.md
- wado-cli/AGENTS.mdAGENTS.md
- wado-compiler/AGENTS.mdAGENTS.md
- wado-from-idl/AGENTS.mdAGENTS.md
- wado-host/AGENTS.mdAGENTS.md
- wado-lsp/AGENTS.mdAGENTS.md
- wado-manifest/AGENTS.mdAGENTS.md
- wado-run-pjrt/AGENTS.mdAGENTS.md
- wado-run-webgpu/AGENTS.mdAGENTS.md
- wado-vscode/AGENTS.mdAGENTS.md
- benchmarkSkill
- ci-timingSkill
- code-review-responseSkill
- coverage-investigationSkill
- crSkill
- debuggerSkill
- eddySkill
- git-upstream-syncSkill
- jcoSkill
- markdownSkill
- on-task-doneSkill
- optimizer-debugSkill
- profiling-wado-compilerSkill
- pull-requestSkill
- rust-renameSkill
- rustSkill
- vendor-submodulesSkill
- wado-cliSkill
- wado-performanceSkill
- wasm-optSkill
The file
We split this skill into the rules it gives an agent
1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 72 edited versions we read keep these.
0
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 17
Every file that reports, whatever its type, plus any doc it made stale.
“Common” is measured: kept by at least 6 in 10 of the 72 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Distill
Scope
B=$(git merge-base origin/main HEAD)
scripts/changed-sources.sh # the files, one per line
git diff "$B" --stat -- $(scripts/changed-sources.sh)
git diff "$B" -- $(scripts/changed-sources.sh)Every file that reports, whatever its type, plus any doc it made stale. Whatever
you notice on the way in is in scope too, pre-existing or not. A WEP keeps the
sections docs/AGENTS.md requires.
Generated files are the one exclusion, and the script is what applies it:
.gitattributes marks them linguist-generated, so a new generated path is
excluded the moment it is marked and nothing here lists paths a second time.
A plain git diff buries the branch under them — on a generator change they
outnumber the sources five to one. Read the sources; regenerate the rest.
This is the scope on every run. Distilling again means the whole branch again, never the diff since the last distill: an earlier pass is not a clean bill, and what the code between the two commits made stale is spread across everything the branch touched.
Answering review feedback is one of the times to run it. Such a fix is written to satisfy a reviewer rather than to fit the code, so it arrives with the reviewer's framing in its comments, an explanation of the bug beside the fix, and often a helper the codebase already had.
Rules
Distilling keeps what the code does. The only change of behaviour it makes is a bug fix, and a bug fix starts from a failing test.
Code
- Reuse: don't re-implement what the codebase already has. Grep the shared modules and the files next to the change, and call the existing helper.
- Duplication: one behaviour, one implementation. Hoist the shared part of near-copies behind the difference — a parameter, a closure, an enum. Don't abstract a single use.
- Altitude: a special case layered on shared infrastructure means the fix sits too shallow. Generalize the mechanism instead.
- Dead code: delete what the change left behind.
- Efficiency: cut wasted work — recomputation, repeated I/O, independent work run in sequence. A stored closure pins everything it captured; prefer a struct holding only the fields it needs.
- Contracts, not defences: a function states what it requires and trusts its
callers. Defensive programming is banned. A default or a fallback whose
validity you cannot argue is the smell. It turns a broken call into a wrong
answer that no test will catch. Write
assert!for what the caller owes andunreachable!for the arm that cannot happen, and say which caller establishes it where that is not obvious. A branch that genuinely can happen is control flow and stays; the ban is on inventing an answer for a case you have not shown to be reachable.
Comments
Apply AGENTS.md § General Rules to every comment in scope. A comment saying what the code does is a rename or a decomposition to make. A comment stating an invariant is an assert to write. A comment that carries no information is deleted outright.
Markdown
Apply the markdown skill to every Markdown file in scope.
In an instruction file (an AGENTS.md, a skill), also cut a rule that is
trivially deduced from a principle the file or the root AGENTS.md already
states: detail dilutes the rules that matter. Keep one that fences a principle
against a reading broader than intended.
Cut words, never a fact. A measurement, a name to grep for (a pass, a type, a function), a condition or an exception, and the bug behind a rule are facts, not detail. Before removing a passage, grep for what links to it. Keep a code block runnable on its own: every name it uses defined in it, every command and path real. Run only a block without side effects to check it.
Sweep by shape
A copy does not share a name, so grepping the name finds nothing.
For every predicate or helper the branch adds or moves, grep the tree for what its body looks like rather than what it is called. Two or three tokens of the body carry further than a signature.
Finish the sweep in the same pass. Fixing the sites a reviewer named and calling it a class fix leaves the rest standing, and the next review returns them one at a time.
Cycle
Three passes over that scope; surviving one is no exemption. Stop when a pass finds nothing to cut.
Finish
mise run formatCode edits: run the tests covering what you touched (mise run test,
mise run test-wado). Comment, doc, and Markdown edits alone need only
mise run check, which holds the specification's examples to their fixtures.
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.

