ponytail
Skill2 starsChanged 6 days agoMITView on GitHub
Use for the smallest complete change (ponytail, minder code, kleinste diff).
144 other projects use a version of this, 35 of them edited.
Adopted by
Most recently taken up by
- Anning01/novelvidsedited, first seen Nov 8, 2026
- kortix-ai/sunatheir own version, first seen Nov 8, 2026
- OpenStrap/edgetheir own version, first seen Nov 8, 2026
- TNTcraftHIM/Piiktheir own version, first seen Nov 8, 2026
- gaoyu06/harnesstheir own version, first seen Nov 8, 2026
- n1kben/agentstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use reimagine, herdr-cli 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.
dev-willbird1936Skillreimagine
Rebuild an instruction file for reliable adherence and lower context cost. Use when explicitly asked to reimagine or restructure-and-compress a prompt, skill or rules file.
Also in 3 projects that use this
julweberSkillherdr-cli
Expert guide for herdr, the terminal workspace manager for AI coding agents.
Also in 3 projects that use this
vikas8520-coderSkillcli-backup-sync
Backup and restore Nerve data from the CLI. Trigger incremental snapshots, sync to cloud storage, manage backup schedules, and restore from archive files.
Also in 3 projects that use this
vikas8520-coderSkillcli-batches
Submit and monitor batch inference jobs from the CLI. Upload and manage files for batch processing, retrieve results, and integrate batch pipelines with CI/CD workflows.
Also in 3 projects that use this
- +8 more from vikas8520-coder
Other versions144 projects use a version of this.
Word for word · 0
None found.
Edited · 35
- DietrichGebert/ponytail★ 182k45% same
- graphif/project-graph★ 4.5k45% same
- Cygnux-Labs/Tracekit★ 60245% same
- mxyhi/ok-skills★ 51145% same
- Anning01/novelvids★ 36245% same
- engasnm111/lnwjud★ 20940% same
- KrishP147/skills★ 3745% same
- Djarvur/go-ultimate★ 3145% same
- taljeon/plzdo★ 945% same
- nesnilnehc/ai-cortex★ 842% same
25 more not listed
Wrote their own version · 109
- cheaper-inference/OmniRoute★ 86ksame name
- diegosouzapw/OmniRoute★ 86ksame name
- kortix-ai/suna★ 21ksame name
- kortix-ai/CompanyOS★ 21ksame name
- jeffvli/feishin★ 10ksame name
- pollinations/pollinations★ 5.3ksame name
- TNTcraftHIM/Piik★ 4.1ksame name
- TanStack/ai★ 3.3ksame name
- marmelab/atomic-crm★ 1.4ksame name
- OpenStrap/edge★ 893same name
99 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 repository99 other files it gives its agents.
- docs/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- brainstormingSkill
- claude-md-improverSkill
- dispatching-parallel-agentsSkill
- executing-plansSkill
- finishing-a-development-branchSkill
- forge-a11ySkill
- forge-agent-reportSkill
- forge-agentSkill
- forge-apiSkill
- forge-authSkill
- forge-autoresearchSkill
- forge-botsSkill
- forge-brainstormSkill
- forge-cliSkill
- forge-cmsSkill
- forge-code-reviewSkill
- forge-contextSkill
- forge-councilSkill
- forge-craftSkill
- forge-databaseSkill
- forge-dataSkill
- forge-debugSkill
- forge-deeplearnSkill
- forge-devopsSkill
- forge-docdriftSkill
- forge-doctorSkill
- forge-ecc-adapterSkill
- forge-ecommerceSkill
- forge-electronSkill
- forge-extensionSkill
- forge-figmaSkill
- forge-fullstackSkill
- forge-gameSkill
- forge-graded-verifySkill
- forge-guardianSkill
- forge-heartbeatSkill
- forge-i18n-docsSkill
- forge-intakeSkill
- forge-integrationSkill
- forge-long-form-writingSkill
- forge-mcp-clientsSkill
- forge-memorySkill
- forge-migrationSkill
- forge-mindmapSkill
- forge-mlopsSkill
- forge-mobileSkill
- forge-n8nSkill
- forge-nightshiftSkill
- forge-observabilitySkill
- forge-paymentsSkill
- forge-plan-largeSkill
- forge-prdSkill
- forge-predictionSkill
- forge-projectbrainSkill
- forge-prompt-coachSkill
- forge-qualitySkill
- forge-ragSkill
- forge-registrySkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 5 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 140 edited versions we read keep these.
5
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 9
Self-improvement substrate: before applying this skill, read learnings.md in this skill's own folder and honor its corrections.
- Distinctive: line 33
Read the task and the code it touches.
- Very specific: line 42
Standard library or platform feature?: Use it, unless the project has its own (a house component beats a native widget).
- Distinctive: line 44
An installed dependency?: Use it. Never add a dependency for a few lines.
- Distinctive: line 50
No abstraction, wrapper, option, config, boilerplate or "for later" code nobody asked for.
- Distinctive: line 52
Keep the structure the codebase already has: its layers, interfaces and conventions.
- Distinctive: line 69
If cutting one of these looks tempting, the change is not finished yet.
“Common” is measured: kept by at least 6 in 10 of the 140 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
Ponytail — the smallest change that fully solves it
Self-improvement substrate: before applying this skill, read learnings.md in this skill's own
folder and honor its corrections. After a run that produced a genuine correction, append it there with
a date and real evidence — never invent a lesson that did not happen.
Work like the senior developer who has seen every over-built feature: the best code is code never written. Solve the whole problem with the least new code. Lazy about the solution, never about reading, finishing or checking.
On and off. On when the owner asks for it ("ponytail", "minder code", "kleinste diff", "write less code"); it stays on for the session until "stop ponytail", "normale modus" or "normal mode". Level switch: "ponytail lite", "ponytail full" (default) or "ponytail ultra". Build Boss may also apply the ladder as a lens on any package; it never overrides the work package or the Forge rules below.
Where this skill stops (siblings)
| Question | Skill |
| Review a diff (order, severity, blocking) | forge-code-review — ponytail adds only the lean lens (references/lean-review.md) |
| Clean up after a finished diff | simplify / refactor-clean |
| Short answers in chat (fewer output tokens) | forge-token-economy terse mode |
| Tests first for new behaviour | test-driven-development / forge-testing |
Extra moves while writing (existence reason, understand first, fix the shared root): references/minimal-code-discipline.md.
Before you write
Read the task and the code it touches. List every place the change must reach: callers, tests, fixtures, config, exports, docs that state the behaviour. Check what the change could break: data it could destroy or expose, callers that stop working. That list is the scope. Extra features are not.
The ladder — take the first rung that fully works
- Does it need to exist? Skip features, options and flexibility nobody asked for; name them in one line. A vague ask ("bouw me X") gets the smallest version that does the core job.
- Already in this codebase? A helper, component, service or pattern: use it the way the surrounding code does. Grep before you write.
- Standard library or platform feature? Use it, unless the project has its own (a house component
beats a native widget). On Windows + Node:
node:fs,node:path,node:testbefore a package. - An installed dependency? Use it. Never add a dependency for a few lines.
- One line a reader understands at a glance? One line.
- Otherwise: the minimum code that works.
Rules while building
- Finish every part the task needs, including the callers, tests and fixtures your change breaks.
- No abstraction, wrapper, option, config, boilerplate or "for later" code nobody asked for. Keep values in the form the platform already gives. Deletion beats addition.
- Keep the structure the codebase already has: its layers, interfaces and conventions.
- The shortest working diff wins once you know everything it must touch. A one-liner that needs decoding is not short.
- Comment only the why the code cannot show, in one line.
- Bug fix: grep every caller of the function you touch, then fix the root cause once in the shared
code (see
forge-debug: root cause first, regression test RED then GREEN). - Code you move or merge keeps its error handling and validation.
- Between two options of equal size, take the one that is correct on edge cases.
- A shortcut with a known limit gets a comment:
shortcut: <the limit>, <when to upgrade>. - Tests: new non-trivial logic (a branch, a loop, a parser, money, security, a new script) leaves at least one small real test that fails when it breaks. Where Forge's testing rules ask for more (TDD for new behaviour, a regression test for every bug fix), Forge's rule wins. Lean never means untested.
Never cut
Validation at trust boundaries · error handling that prevents data loss or a silent failure · security (auth, secrets, injection, path checks) · accessibility (labels, focus states, contrast, keyboard use) · the calibration real hardware needs · anything the owner or the work package asked for. If cutting one of these looks tempting, the change is not finished yet.
Levels
| Level | Behaviour |
| lite | Build what was asked. Name the smaller option in one line and let the owner pick. |
| full | The ladder and rules above. Default. |
| ultra | Also question the ask before building: push back on any part the need does not justify, in one short list. Never drop a part on your own; the owner decides. |
End of every reply
One or two plain lines: what you skipped or did not check, and any risk the owner must know. This is the honesty line, not a summary: it may say "none".
Forge fit
- The work package and Head Chef's scope stay the boundary; ponytail makes the diff smaller inside it, it never shrinks the scope.
- GOAT routing, the verifier gate, the Stop test gate and the hard gates are unchanged.
- Ponytail is about code size. It does not make reports, QA verdicts or copy-paste prompts shorter.
Provenance
- DietrichGebert/ponytail — paths
.openclaw/skills/ponytail/SKILL.mdand.openclaw/skills/ponytail-review/SKILL.md,README.md(doctrine only); pinned SHA96608e078b2d5ea607c5c3b8b5a54d1c1fba6846; MIT License, Copyright (c) 2026 DietrichGebert. Taken (rewritten for Forge Bosses, Windows + Node and GOAT): the "write less code, never cut validation, error handling, security or accessibility" doctrine, the ladder, the build rules, the shortcut comment, the end-of-reply risk line, the lite/full/ultra levels, and the lean review categories (inreferences/lean-review.md). Excluded: all Node hooks and plugins (.opencode/plugins/*, frontmatter hook), every command.tomland per-tool rules copy (commands/,.cursor,.windsurf,.kiro, …), the "band" status bar and "board" local page (they write under the user's home folder), the audit, debt, gain and help skills, install steps, and the README benchmark numbers (not measured here). The MIT notice above applies to the adapted portions; the full MIT text is at the source repository. Verified 2026-11-05: the three paths above and the rootLICENSE("Copyright (c) 2026 DietrichGebert") were fetched read-only at the pinned SHA from raw.githubusercontent.com (HTTP 200; the GitHub API was unreachable); noNOTICEfile exists at that SHA. Hidden-Unicode scan of the fetched text: clean.
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.

