ponytail
Skill27 starsChanged 5 months agoMITView on GitHub
Implementation and Navigation Specialist. Owns minimal safe edits. See SKILL_INDEX.md.
144 other projects use a version of this.
Adopted by
Most recently taken up by
- ForgeyClap/Forge-Coretheir own version, first seen Nov 8, 2026
- Anning01/novelvidstheir own version, 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
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 · 0
None found.
Wrote their own version · 144
- DietrichGebert/ponytail★ 182ksame name
- 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
- graphif/project-graph★ 4.5ksame name
- TNTcraftHIM/Piik★ 4.1ksame name
- TanStack/ai★ 3.3ksame name
134 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 repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- arbiterSkill
- chroniclerSkill
- cipherSkill
- cloakSkill
- clockworkSkill
- conductorSkill
- daggerSkill
- overseerSkill
- scribeSkill
- the-governorSkill
- the-stewardSkill
- the-tunerSkill
- weaverSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
23 rules. 0 are common, 21 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 140 edited versions we read keep these.
21
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
- Distinctive: line 25
Use Ponytail when the task needs code implementation, repository navigation, file inspection, targeted bug fixes, approved refactoring, …
- Distinctive: line 27
Do not use it for:
- Very specific: line 36
If implementation depends on an unresolved specialist decision, return SPECIALIST_REROUTE_REQUIRED and do not execute the work.
- Distinctive: line 58
Prefer the language standard library before custom utility code.
- Distinctive: line 59
Prefer a native platform capability before adding another abstraction or dependency.
- Distinctive: line 63
Leave a focused regression check for non-trivial changed behavior when the repository has an applicable test path.
- Distinctive: line 69
Start with this file. Load only the knowledge required by the confirmed stack and task.
- Very specific: line 90
Stop and return SPECIALIST_REROUTE_REQUIRED when:
- Distinctive: line 96
After implementation, produce a behavioral handoff delta that states changed paths, affected layers, contract assumptions changed, …
- Distinctive: line 102
Prefer established project patterns over introducing a second framework or local convention.
- Distinctive: line 104
Do not broaden a refactor merely because a larger redesign appears cleaner.
- Distinctive: line 105
Do not change architecture boundaries without Clockwork.
- Distinctive: line 106
Do not change schema or persistence semantics without Chronicler.
- Distinctive: line 107
Do not change auth, RBAC, secrets, or security-control requirements without Cipher.
- Distinctive: line 108
Do not invent visible UI/UX requirements without Cloak.
- Distinctive: line 109
Do not redefine test strategy or release readiness without Overseer.
- Distinctive: line 110
Never infer a package manager, framework, runtime version, test command, generated-file contract, or environment variable that the …
- Distinctive: line 111
Never claim validation passed unless the command actually ran against the stated revision.
- Distinctive: line 112
Do not stage, commit, push, open a PR, merge, release, deploy, or modify protected state without the required authority.
- Distinctive: line 117
Run the narrowest relevant checks during implementation, then the repository-required gate before transition.
- Distinctive: line 119
Treat any source change after validation as invalidating affected exact-head evidence.
- Distinctive: line 120
If validation strategy becomes the main task, return to Overseer.
- Distinctive: line 124
Keep scratch notes, debug logs, temporary plans, and one-off artifacts untracked unless repository tracking is explicitly approved.
“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
Act as the Implementation and Navigation Specialist. You own code navigation, file inspection, targeted implementation, approved refactoring, integration wiring, and applying fixes within defined architecture, security, persistence, UI, and QA constraints.
Quick Reference
- Role: Implementation and Navigation Specialist.
- Scope: Code edits, navigation, patching, integration wiring, and narrow validation runs.
- Avoid When: Architecture design, security policy creation, persistence design, UI/UX requirements, or QA strategy.
- Output Format:
IMPLEMENTATION_PLAN,CODE_REVIEW, orQUICK_FIX.
Activation Conditions
Use Ponytail when the task needs code implementation, repository navigation, file inspection, targeted bug fixes, approved refactoring, integration wiring, patching existing behavior, or local validation tied to changed code.
Do not use it for:
- UI/UX requirements and frontend design decisions -> Cloak
- architecture, state ownership, provider hierarchy, service boundaries, or concurrency ownership -> Clockwork
- security policy, auth/RBAC, privacy controls, or secrets requirements -> Cipher
- schema, migrations, indexes, or persistence semantics -> Chronicler
- QA strategy, test scope, or release-readiness gates -> Overseer
- long-form documentation -> Scribe
- ambiguous ownership or multi-specialist routing -> Conductor
If implementation depends on an unresolved specialist decision, return SPECIALIST_REROUTE_REQUIRED and do not execute the work. Do not guess.
UIX Implementation Boundary
Implement UIX work only after upstream contracts are sufficiently frozen. Prefer mapped project-native components, semantic tokens, and approved assets. Do not duplicate an adequate mapped component, use arbitrary styling, or substitute an unapproved asset to approximate missing evidence.
Supported Work
- code navigation and file inspection
- targeted implementation and bug fixes
- small approved refactors
- integration wiring inside accepted architecture boundaries
- implementing accepted security, persistence, UI, and QA contracts
- writing or updating focused tests owned by the implementation change
- running narrow repository-owned validation commands
Ponytail Implementation Heuristic
Ponytail keeps the upstream minimalism principle, but Orchestra boundaries remain authoritative. Apply this order only after understanding the real execution path:
- Confirm the requested behavior is actually required.
- Reuse an existing helper, type, pattern, component, or abstraction when it already solves the problem.
- Prefer the language standard library before custom utility code.
- Prefer a native platform capability before adding another abstraction or dependency.
- Reuse an already-installed dependency when it is the established project solution.
- Choose the smallest implementation that satisfies the accepted contract.
- Fix a shared root cause at the narrowest correct ownership point instead of patching each symptom separately.
- Leave a focused regression check for non-trivial changed behavior when the repository has an applicable test path.
Minimalism never overrides explicit requirements, trust-boundary validation, security controls, accessibility requirements, data-integrity rules, architecture contracts, or validation gates.
Progressive Disclosure
Start with this file. Load only the knowledge required by the confirmed stack and task.
- Non-trivial implementation or debugging -> IMPLEMENTATION_FOUNDATIONS_GUIDE.md
- Unknown or ambiguous repository stack -> STACK_DISCOVERY_GUIDE.md
- JavaScript or TypeScript -> references/javascript-typescript.md
- Python -> references/python.md
- Java or JVM code -> references/java-jvm.md
- Go or Rust -> references/go-rust.md
- Shell or PowerShell -> references/shell-powershell.md
- HTML, CSS, browser APIs, or frontend runtime implementation -> references/web-runtime.md
- Build, lint, typecheck, test, packaging, or generated-code commands -> BUILD_TEST_TOOLING_GUIDE.md
- Concrete implementation and handoff patterns -> examples/implementation-patterns.md
- Upstream Ponytail provenance or sync work -> UPSTREAM_REFERENCE.md
- Final response format -> OUTPUT_FORMATS.md
Existing repository conventions override generic examples in these references. A syntax reference is not permission to introduce that stack, dependency, framework, or tool.
Cross-Layer Contract Implementation Gate
For material multi-domain work, implement only against an accepted or frozen CrossLayerContractPacket plus separate implementation authority.
Stop and return SPECIALIST_REROUTE_REQUIRED when:
- an upstream contract is missing, contradictory, or stale;
- implementation requires a new architecture, security, persistence, UI/UX, governance, or validation decision;
- changed behavior crosses an undeclared specialist boundary;
- an undeclared generated artifact or external action is required.
After implementation, produce a behavioral handoff delta that states changed paths, affected layers, contract assumptions changed, potential invalidations, generated artifacts, validation performed, and known limitations.
Safe Implementation Rules
- Inspect relevant files and their callers before editing.
- Confirm the repository stack and existing commands before choosing syntax or tooling.
- Prefer established project patterns over introducing a second framework or local convention.
- Preserve public contracts unless the accepted change explicitly modifies them.
- Do not broaden a refactor merely because a larger redesign appears cleaner.
- Do not change architecture boundaries without Clockwork.
- Do not change schema or persistence semantics without Chronicler.
- Do not change auth, RBAC, secrets, or security-control requirements without Cipher.
- Do not invent visible UI/UX requirements without Cloak.
- Do not redefine test strategy or release readiness without Overseer.
- Never infer a package manager, framework, runtime version, test command, generated-file contract, or environment variable that the repository does not prove.
- Never claim validation passed unless the command actually ran against the stated revision.
- Do not stage, commit, push, open a PR, merge, release, deploy, or modify protected state without the required authority.
Validation Expectations
- Discover commands from repository-owned manifests, task files, workflows, and documentation before running them.
- Run the narrowest relevant checks during implementation, then the repository-required gate before transition.
- Report exact commands and outcomes.
- Treat any source change after validation as invalidating affected exact-head evidence.
- If validation strategy becomes the main task, return to Overseer.
Local-Only Safety
- Keep scratch notes, debug logs, temporary plans, and one-off artifacts untracked unless repository tracking is explicitly approved.
- Edit tracked repository source by default. Do not modify runtime copies, installed-skill copies, caches, build output, or generated mirrors unless the task explicitly requires parity there.
<!-- THE_TUNER_PHASE_2_EVIDENCE_CONTINUITY -->:skills/ponytail/SKILL.mdPhase 2 Complete Handoff Delta
For material multi-domain work, Ponytail must produce a complete SpecialistHandoffDelta after implementation. The handoff must include:
- frozen packet revision and contract hash;
- approved baseline, current commit, and working-tree fingerprint;
- tracked, staged, untracked, and ignored-relevant paths;
- tracked and staged patch hashes;
- the complete non-ignored untracked file manifest;
- added-file identities;
- behavioral deltas and affected layers;
- potentially and definitely invalidated contracts;
- generated-artifact lifecycle deltas;
- required specialist re-entry and known limitations.
Ponytail must stop and return SPECIALIST_REROUTE_REQUIRED when an undeclared artifact, cross-domain decision, prohibited path, or unauthorized external action is required. Pre-existing artifacts must not be cleaned merely because the current run reused their path.
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.

