language setting

Global only, ignored in skill frontmatter.

Across the graph 4 layers

One node in a single signed graph. Here is how this primitive connects across the other layers.

Composed in · workflow recipes 1 of the corpus

Multi-Language Documentation GeneratorLanguage setting changes Claude's response language. Fork isolates each language context. Together = parallel generation in multiple languages. (BLOCKED - language is global not)

The language setting configures Claude Code's response language globally via settings.json, not per-task or skill.

What It Is

language is a session-wide configuration field intended to direct Claude Code to respond in a specified language (e.g., "japanese"). Per the feature changelog, it operates at the global settings level (~/.claude/settings.json) and applies to all subsequent Claude responses, rather than being a per-skill or per-task control.

Test Findings

Testing (2026-01-07) attempted two approaches:

Skill frontmatter placement (FAILED) - A skill SKILL.md with language: japanese in the frontmatter produced English responses. The runtime ignored the field entirely.
Global setting hypothesis (UNTESTED) - The global settings.json mechanism was not empirically validated. If implemented, it would require adding "language": "japanese" to ~/.claude/settings.json and restarting Claude Code to take effect.

Result: PARTIAL / INCONCLUSIVE. Skill-level language switching does not work. Global application remains unverified by runtime test.

Scope and Trade-offs

Language is session-wide, not granular. This design rules out per-skill language switching within a single session. Multi-language workflows would require either prompt-based language instructions embedded in task text, or separate Claude Code instances with different global language settings. The missing global test leaves the actual mechanism unconfirmed.

Status

Empirically tested: skill frontmatter does not apply language setting. Claimed mechanism: global settings.json. Verification: incomplete. Operator can resolve this by testing global application directly or consulting Claude Code release notes for language feature scope.

Evidence & receipt

commit98993e4 ↗

◇ ed25519 receipt

idprimitive_aae92562fe63f02cc4c66571

alged25519

pubkey9b87705613b1e2fd064d57fa75a6b679d2856ceafad6b1daa8f982493871b6dd

sig

bde85fa56dfcc5421642496838480358ede9fa58341b2d68d78b8407988b6b278f84d185239a5978ef345c098591d88541a4e51f61c835da7147c6b87a307b0f

Signed with an ed25519 key held off the repo. Anyone can verify against the published public key; nobody without the secret key can forge it. Click verify: it recomputes the signature in your browser. The signature proves integrity and authorship of this exact content — not a third-party timestamp or that the underlying claim is objectively true. signedAt is when the @f3/attest pipeline ran, not when the work happened; the evidence refs carry the source dates.

Connected

instance-of Skill Runtime Class
introduces (in) Claude Code 2.1.0 (Night Zero) Release
verifies (in) language setting — runtime test Test

language setting

Primitive

Kind	Primitive
Introduced	Claude Code 2.1.0
Class	Skill Runtime
Epistemic	inconclusive · runtime-test
Evidence	1 ref
Composed in	1 workflow recipe
Receipt	ed25519 · verifiable

Hands-on runtime test · Night Zero, 2026-01-07

Cite this page

Hunter, R. "language setting." The Runtime Atlas. Content-addressed entity primitive_aae92562fe63f02cc4c66571, ed25519-signed (integrity, not third-party timestamp). https://ryanhunter.io/atlas/e/language-setting

@misc{atlas-language-setting,
  title  = {language setting},
  author = {Hunter, Ryan},
  note   = {The Runtime Atlas; entity primitive_aae92562fe63f02cc4c66571; ed25519-signed for integrity},
  url    = {https://ryanhunter.io/atlas/e/language-setting}
}