ywc-setup-language
(ywc) Set or inspect the persistent output language for ywc-generated documents, PRs, and commit messages.
- License 许可
- MIT
- 最近更新
- 2026-11-08
- Stars
- ★ 8
- Forks
- 1
01 前言 Front Matter
- name
- ywc-setup-language
- phase
- setup
- version
- 1.0.0
- category
- config
- requires
- []
- description
- (ywc) Set or inspect the persistent output language for ywc-generated documents, PRs, and commit messages.
- allowed tools
- Read, Edit, Write, Bash, Grep, Glob
- advisor_budget
- 0
- disable-model-invocation
- true
02 资源文件 6
- 参考文档README.en.md2.5 KB
- 参考文档README.es.md2.5 KB
- 参考文档README.ja.md2.6 KB
- 参考文档README.ko.md2.5 KB
- 参考文档README.md2.5 KB
- 参考文档README.zh.md2.5 KB
03 SKILL.md 正文
Setup Language Skill
Announce at start: "I'm using the ywc-setup-language skill to persist the output language into the appropriate CLAUDE.md."
Persist a project- or user-level output language into a canonical ## Language Policy
section of the appropriate CLAUDE.md, so every language-aware ywc-* skill produces
documents, PR text, and commit messages in that language with no per-call --lang flag and
no per-call prompt. Reading the policy back is a separate --show mode.
This skill only writes the policy. How consuming skills read and resolve it (the precedence chain, code list, and section format) is defined once elsewhere:
After normalizing the input (below), run
bash "${CLAUDE_SKILL_DIR}/../scripts/resolve-language.sh" --emit-section <code>. It prints the canonical## Language Policysection body (with<code>substituted) — write it verbatim into the targetCLAUDE.md.
Rationalization Defense
When tempted to skip a rule, check this table first:
| Excuse | Reality |
|---|---|
| "The file already has some language note, I'll just append a new section" | Appending creates a duplicate ## Language Policy heading (AC3 violation). Setup is a create-or-replace on the delimited section — exactly one such heading must exist afterward. |
| "There's an ad-hoc language sentence elsewhere in CLAUDE.md, I'll rewrite it to match" | Out of scope. Setup writes only the canonical delimited section; it never migrates or rewrites pre-existing prose. |
"--user target ~/.claude/CLAUDE.md doesn't exist, so this is an error" |
Not an error. Create the file with only the delimited section (EC5) — do not fabricate other global instructions. |
"The user typed korean, that's not a valid code, stop" |
Normalize full names to codes (korean→ko, EC3) before writing. Reject only inputs that map to none of ko|ja|en|es|zh. |
"--show found no policy, I should write a default so future calls work" |
--show is read-only (AC4). Report "no policy configured" and the fallback source — never write during show. |
| "I'll add a forced global default so absence resolves cleanly" | No. The no-block invariant (NFR1) requires absence to fall through to each consuming skill's own fallback. Setup never introduces a global default. |
Arguments
| Argument | Form | Example | Description |
|---|---|---|---|
| Language | positional | ko, japanese |
Output language code (ko|ja|en|es|zh) or full name (korean, japanese, english, spanish, chinese). Normalized to a code before writing. Required unless --show. |
--user |
flag | Target the user-global ~/.claude/CLAUDE.md instead of the project CLAUDE.md (the default target). |
|
--show |
flag | Read-only: report the resolved language and its source rung (project / user / none). No write. |
Procedure
Write mode (default — positional language given)
- Normalize the positional argument to a code using the mapping in
references/language-resolution.md. If it maps to none ofko|ja|en|es|zh, stop and report the accepted values — do not write. - Select the target file: project
CLAUDE.md(repo root) by default, or~/.claude/CLAUDE.mdwhen--useris set. - Create-or-replace the
## Language Policysection (idempotent):- If the target file has a
## Language Policysection, replace that delimited section in place — its heading through the line before the next##heading (or EOF). Exactly one## Language Policyheading must remain afterward (AC3, EC6). - If the section is absent, append the canonical section.
- If the target file itself is absent (common on
--user), create it containing only the canonical section (EC5) — no other global instructions. - The section body is the canonical format from
references/language-resolution.mdwith the resolved code filled into**Output language**.
- If the target file has a
- Confirm with a one-line message naming the written file path and code (no
@-activation prompt — user-global~/.claude/CLAUDE.mdis already auto-loaded).
Show mode (--show)
- Resolve per the precedence chain in
references/language-resolution.md(project policy → user policy → none), reading eachCLAUDE.md's## Language Policysection. - Report the resolved code and the winning source rung, or "no policy configured" with the fallback source when neither file has a valid section. A malformed section is reported as invalid (EC2), not silently ignored.
- Never write in show mode.
Notes
- The canonical
## Language Policysection supersedes older CLAUDE.md language cues; when it is absent, consuming skills still honor their own prior fallback (see the reference). This preserves no-regression behavior for projects that have not yet run this skill. - Keep-in-English tokens (conventional-commit
type:prefix, PR-title[task-id]/prefix, technical terms) are unaffected by the configured language — they stay English by policy.
04 同仓库更多 Skills
missing-frontmatter
暂无简介
malformed-frontmatter
暂无简介
fixture-501-fail
(ywc) Use when aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa Do not use for x.
fixture-500-pass
(ywc) Use when aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa Do not use for x.
ywc-worktrees
(ywc) Use when creating, auditing, pruning, or resolving git worktrees for isolated task workspaces, whether delegated from an executor or used standalone. Triggers: "worktree 생성", "worktree audit", "worktree 정리", "worktree prune", "ywc-worktrees", "create worktree". Do not use for branch creation/deletion alone, PR creation (use ywc-create-pr), CI or merge delivery (use ywc-finish-branch), task orchestration (use ywc-sequential-executor or ywc-parallel-executor), or non-git isolation.
ywc-wayfinder
(ywc) Use when a large or uncertain change needs multi-session discovery with a deterministic local map and exactly one active ticket before planning or implementation. Triggers: "wayfinder", "길찾기", "discovery map", "multi-session discovery", "탐색 ticket", "経路整理", "探索マップ", "ywc-wayfinder". Do not use for ordinary planning (use ywc-plan), idea clarification (use ywc-brainstorm), direct implementation, or tracker/database writes outside the local Markdown map.