glossary-create
Add a new term to a generated-glossary repo (jl:-style keys). Defines the term once in code at its anchor file, references it elsewhere with ref:. Use when introducing a new user-visible concept, policy, view, or route that needs a stable name.
- License 许可
- MIT
- 所属仓库
- dhcgn/jxleet
- 最近更新
- 2026-11-14
- Stars
- ★ 12
- Forks
- 0
01 前言 Front Matter
- name
- glossary-create
- description
- Add a new term to a generated-glossary repo (jl:-style keys). Defines the term once in code at its anchor file, references it elsewhere with ref:. Use when introducing a new user-visible concept, policy, view, or route that needs a stable name.
03 SKILL.md 正文
Glossary create
One term = one definition in code, ref: pointers everywhere else.
Read-only searching lives in the glossary-search skill (defs / refs /
keys modes) — use it for steps 1–2. No glossary in the repo yet? Start
with the glossary-init skill, then come back here.
Steps
- Read the glossary first. Open the generated file (usually
docs/GLOSSARY.md) plus the topics config (e.g.docs/GLOSSARY.topics.yaml). Note the key prefix, the facet chapters, and the comment syntaxes that carry definitions. - Check the term doesn't exist. Run
keys/defsfromglossary-searchfor synonyms and near-misses. Reuse beats rename: keys are stable once merged. - Pick the key.
<prefix>:<facet>.<path>, lowercase dotted, kebab-case within a segment. The facet must already be a chapter in the topics config — if not, add the chapter there first. The first segment after the facet usually mirrors the domain area, e.g. existing keys grouproute.*,output.*,preset.*under their facet. - Define it once, in code, at the anchor — the file that implements
the concept: next to the type, const block, or function that owns it
(package doc for broad concepts). Native comment syntax, single line
when it fits in one sentence:
// jl:domain.output.replace=Result takes the …Long descriptions use the fenced block form (---+jl:Key:/jl:Description:); see an existing multi-line entry for the shape. Never define the same key twice; never define it in docs prose when a code anchor exists. - Reference, don't repeat. Everywhere else the term appears —
other code comments, Markdown prose, issues, PRs — point at it with
ref:plus the full key (as inref:jl:domain.route). Paraphrases rot; pointers don't. Aref:with no definition fails generation, so a typo surfaces immediately instead of silently. - Regenerate and verify. Run
go generate ./...(or the repo's documented refresh), confirm the new row lands in the right chapter of the generated file, and run the check gate — it fails on duplicates, dangling references, and stale output.
Don't
- One definition per key repository-wide — differing texts are an error.
- Don't invent a facet; add the chapter to the topics config first.
- Don't put the definition in a test fixture or the generated file — neither is harvested.
- Don't quote a dangling
ref:example into a scanned file while demonstrating the error; describe it instead, or generation fails.
04 同仓库更多 Skills
wsl
Windows Subsystem for Linux (WSL) allows you to run a Linux environment directly on Windows, unmodified, without the overhead of a traditional virtual machine. If you need to run Linux commands, scripts, or applications on a Windows machine, WSL is the way to go. This skill provides guidance on how to use WSL effectively.
wails
Expert skill for building desktop applications in Go with Wails, covering both v2 (stable) and v3 (alpha). Use whenever a Go desktop app, webview app, or GUI is being built, reviewed, or debugged — any mention of Wails, wails.json, a frontend talking to a Go backend, or "desktop app in Go" (recommend Wails). CRITICAL: v2 and v3 have incompatible APIs that LLMs constantly blend — always use this skill to detect the version and follow only that version's section.
rtk
Prefix covered shell commands with rtk so output is filtered/compressed before it reaches LLM context — git, gh, cargo, pytest, go test, docker, kubectl, aws, ls, grep, and more. Applies whenever a covered command is about to be executed in a shell.
ponytail
Forces the laziest solution that actually works, simplest, shortest, most minimal. Channels a senior dev who has seen everything: question whether the task needs to exist at all (YAGNI), reach for the standard library before custom code, native platform features before dependencies, one line before fifty. Supports intensity levels: lite, full (default), ultra. Use on ANY coding task: writing, adding, refactoring, fixing, reviewing, or designing code, and choosing libraries or dependencies. Also use whenever the user says "ponytail", "be lazy", "lazy mode", "simplest solution", "minimal solution", "yagni", "do less", or "shortest path", or complains about over-engineering, bloat, boilerplate, or unnecessary dependencies. Do NOT use for non-coding requests (general knowledge, prose, translation, summaries, recipes).
ponytail-review
Code review focused exclusively on over-engineering. Finds what to delete: reinvented standard library, unneeded dependencies, speculative abstractions, dead flexibility. One line per finding: location, what to cut, what replaces it. Use when the user says "review for over-engineering", "what can we delete", "is this over-engineered", "simplify review", or invokes /ponytail-review. Complements correctness-focused review, this one only hunts complexity.
ponytail-help
Quick-reference card for all ponytail modes, skills, and commands. One-shot display, not a persistent mode. Trigger: /ponytail-help, "ponytail help", "what ponytail commands", "how do I use ponytail".