Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Translate Content

ASecurity

Translate agent-almanac content (skills, agents, teams, guides) into a target locale while preserving code blocks, IDs, and technical structure. Covers scaffolding, frontmatter setup, prose translation, code preservation, and freshness tracking. Use when localizing content for a new language, updating stale translations after source changes, or batch-translating a domain.

31 stars
0 votes
0 copies
0 views
Added 9/3/2026
ai-agentsbashnodegit

Security Analysis

A100/100

Scanned 9/3/2026

$npx -y skills add pjt222/agent-almanac --skill translate-content --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Translate Content?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Translate Content
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/pjt222-translate-content-8e1f7121/badge)](https://www.skillsdirectory.com/skills/pjt222-translate-content-8e1f7121)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: translate-content
description: >
  Translate agent-almanac content (skills, agents, teams, guides) into a target
  locale while preserving code blocks, IDs, and technical structure. Covers
  scaffolding, frontmatter setup, prose translation, code preservation, and
  freshness tracking. Use when localizing content for a new language, updating
  stale translations after source changes, or batch-translating a domain.
license: MIT
allowed-tools: Read Write Edit Bash Grep Glob
metadata:
  author: Philipp Thoss
  version: "1.1"
  domain: i18n
  complexity: intermediate
  language: multi
  tags: i18n, translation, localization, multilingual, l10n
---

# Translate Content

Translate English source content into a target locale, preserving technical accuracy and structural integrity.

## When to Use

- Localizing a skill, agent, team, or guide into a supported language
- Updating a translation that has become stale after source changes
- Batch-translating multiple items within a domain or content type
- Creating initial translations for a new locale

## Inputs

- **Required**: Content type — `skills`, `agents`, `teams`, or `guides`
- **Required**: Item ID — the name/identifier of the content (e.g., `create-r-package`)
- **Required**: Target locale — IETF BCP 47 code (e.g., `de`, `zh-CN`, `ja`, `es`)
- **Optional**: Batch list — multiple IDs to translate in sequence

## Procedure

### Step 1: Read the English source

1.1. Determine the source file path:
   - Skills: `skills/<id>/SKILL.md`
   - Agents: `agents/<id>.md`
   - Teams: `teams/<id>.md`
   - Guides: `guides/<id>.md`

1.2. Read the entire source file to understand context, structure, and content.

1.3. Identify sections that must stay in English:
   - All code blocks (fenced with triple backticks)
   - Inline code (backtick-wrapped)
   - YAML frontmatter field names and technical values (`name`, `tools`, `model`, `priority`, `skills` list entries, `allowed-tools`, `tags`, `domain`, `language`)
   - File paths, URLs, command examples
   - `<!-- CONFIG:START -->` / `<!-- CONFIG:END -->` blocks in teams

**Expected:** Full understanding of source content with clear mental separation of translatable prose vs preserved technical content.

**On failure:** If source file is not found, verify the ID exists in the registry. Check for typos in the content type or ID.

### Step 2: Scaffold the translation file

2.1. Run the scaffolding script:
```bash
npm run translate:scaffold -- <content-type> <id> <locale>
```

2.2. If the file already exists, read it to check whether it needs updating (stale) or is already current.

2.3. Verify the scaffolded file has translation frontmatter fields:
   - `locale` — matches target locale
   - `source_locale` — `en`
   - `source_commit` — current git short hash; the English revision a human translated against
   - `fence_basis_commit` — the revision this file's frozen fences were verified against (#552).
     Equal to `source_commit` on a fresh scaffold, because a scaffold is a byte copy, and they
     diverge from the first edit of either kind. Never move this one by hand.
   - `translator` — attribution string
   - `translation_date` — today's date

**Expected:** Scaffolded file at `i18n/<locale>/<content-type>/<id>/SKILL.md` (or `.md` for other types) with correct frontmatter.

**On failure:** If the scaffold script fails, create the directory manually with `mkdir -p` and copy the source file. Add frontmatter fields manually.

### Step 3: Translate the description

3.1. Translate the `description` field in the YAML frontmatter into the target locale.

3.2. For skills, the description is inside the top-level frontmatter. For agents/teams/guides, it is also in the top-level frontmatter.

3.3. Keep the translation concise — match the length and style of the original.

**Expected:** Description field contains an idiomatic translation that accurately conveys the original meaning.

**On failure:** If the description is ambiguous, keep it closer to literal translation rather than risk misinterpretation.

### Step 4: Translate prose sections

4.1. Translate all prose content section by section:
   - Section headings (e.g., "## When to Use" → "## Wann verwenden" in German)
   - Paragraph text
   - List item text (but not list item code/paths)
   - Table cell text (but not table cell code/values)

4.2. Preserve these elements exactly as-is:
   - Code blocks (``` fenced and indented)
   - Inline code (`backtick-wrapped`)
   - File paths and URLs
   - Skill/agent/team IDs in cross-references
   - YAML/JSON configuration examples
   - Command-line examples
   - `**Expected:**` and `**On failure:**` markers (translate the label, keep the structure)

4.3. For skills, translate the standardized section names:
   - "When to Use" → locale equivalent
   - "Inputs" → locale equivalent
   - "Procedure" → locale equivalent
   - "Validation" → locale equivalent
   - "Common Pitfalls" → locale equivalent
   - "Related Skills" → locale equivalent

4.4. For agents, translate:
   - Purpose, Capabilities, Available Skills (section name only — skill IDs stay English), Usage Scenarios, Best Practices, Examples, Limitations, See Also

4.5. For teams, translate:
   - Purpose, Team Composition (prose only — IDs stay English), Coordination Pattern, Task Decomposition, Usage Scenarios, Limitations

4.6. For guides, translate:
   - All prose sections, troubleshooting text, table descriptions
   - Keep command examples, code blocks, and configuration snippets in English

**Expected:** All prose sections translated idiomatically. Code blocks identical to English source. Cross-references use English IDs.

**On failure:** If uncertain about a technical term, keep the English term with a parenthetical translation. Example: "Staging-Bereich (Staging Area)" in German.

### Step 5: Verify structural integrity

5.1. Confirm the translated file has the same number of sections as the source.

5.2. For skills, verify all required sections are present:
   - YAML frontmatter with `name`, `description`, `allowed-tools`, `metadata`
   - When to Use, Inputs, Procedure, Validation, Common Pitfalls, Related Skills

5.3. Verify code blocks are identical to the English source by running the checker,
   not by reading:

   ```bash
   node scripts/check-i18n-fence-parity.js --locale <locale> --id <content-id>
   ```

   Scope it to the file you just wrote. `npm run validate:i18n-fences` runs the
   whole corpus, which still carries a backlog (#477) and would fail for reasons
   unrelated to your work — training exactly the ignore-the-red-check reflex
   this gate exists to prevent.

   Every fence whose tag is not `text`, `markdown` or `md` is frozen: its body must
   match the English source byte-for-byte, comments and string literals included.
   Repair with `npm run normalize:i18n-fences`, which previews; add `-- --write`
   to apply once the preview looks right. `-- --tag yaml,json` scopes the run to
   given fence tags, which is how the #477 backlog is cleared in reviewable
   batches.

   The normalizer covers all four content trees — `skills`, `agents`, `teams`,
   `guides` — so it repairs exactly what the checker flags. `-- --tree
   guides,agents` scopes a run by tree the way `--tag` scopes it by fence tag,
   which is how the mirror slice was cleared as its own batch (#518). A `--tree`
   naming a tree your `--locale` carries no translations for exits 2 rather than
   reporting a clean-looking zero.

   This step said "diff the fenced blocks" from the day the i18n tree was created,
   and the corpus still accumulated 1,307 violations, because reading a fence and
   diffing it are not the same act (#472). Run the command.

5.4. Check line count: skills must be ≤ 500 lines.

5.5. Verify `name` field matches the English source exactly (it is the ID, never translated).

**Expected:** Structurally valid translated file that passes validation.

**On failure:** Compare section-by-section with the English source. Restore any missing sections.

### Step 5.5: Verify prose is translated

5.5.1. Sample 3 prose paragraphs from the body of the translated file. Choose paragraphs from different sections — not headings, not code blocks, not frontmatter.

5.5.2. Confirm each sampled paragraph is written in the target language, not English.

5.5.3. If any sampled paragraph is still in English, the translation is incomplete. Return to Step 4 and translate the remaining English prose before proceeding.

**Expected:** All 3 sampled prose paragraphs are in the target language, confirming the body text has been translated — not just headings and frontmatter.

**On failure:** Identify which sections still contain English prose. Translate them before continuing to Step 6.

### Step 6: Write the translated file

6.1. Write the complete translated content to the target path using the Write or Edit tool.

6.2. Verify the file exists at the expected path:
   - Skills: `i18n/<locale>/skills/<id>/SKILL.md`
   - Agents: `i18n/<locale>/agents/<id>.md`
   - Teams: `i18n/<locale>/teams/<id>.md`
   - Guides: `i18n/<locale>/guides/<id>.md`

**Expected:** Translated file written to disk at the correct path.

**On failure:** Check directory exists. Create with `mkdir -p` if needed.

## Validation

- [ ] Translated file exists at `i18n/<locale>/<type>/<id>`
- [ ] `name` field matches English source exactly
- [ ] `locale` field matches target locale
- [ ] `source_commit` field is set to a valid git short hash
- [ ] `check-i18n-fence-parity.js --locale <locale> --id <id>` passes for this file
- [ ] All cross-referenced IDs (skills, agents, teams) are in English
- [ ] File is under 500 lines (for skills)
- [ ] `npm run validate:translations` reports no issues for this file
- [ ] Prose reads idiomatically in the target language

## Common Pitfalls

- **Translating code blocks**: Code, commands, and configuration must stay in English. Only translate surrounding prose.
- **Translating the `name` field**: The `name` field is the canonical ID. Never translate it.
- **Translating tag values**: Tags in `metadata.tags` stay in English for cross-locale consistency.
- **Inconsistent terminology**: Use the same translation for a technical term throughout the file and across files in the same locale.
- **Literal translation of idioms**: Translate the meaning, not the words. "Common Pitfalls" should become the locale's natural equivalent, not a word-for-word translation.
- **Missing `source_commit`**: Without this field, freshness tracking breaks. Always include it.
- **Batch throughput over quality**: Scaffolding-only output — where headings are translated but body text remains in English — is not a valid translation. Prefer fewer complete translations over many partial ones.
- **Exceeding 500 lines**: Translations may expand ~10-20% vs English. If near the limit, tighten prose rather than removing content.

## Related Skills

- [create-skill](../create-skill/SKILL.md) — understand the SKILL.md structure being translated
- [review-skill-format](../review-skill-format/SKILL.md) — validate translated skill structure
- [evolve-skill](../evolve-skill/SKILL.md) — update skills that have changed since translation

Attribution

pjt222pjt222
View sourceSee grades on GitHubMore from pjt222 →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698461 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →