.llm

# DialectOS — LLM Crawler Optimized Summary
# Format: https://llm.txt (proposed standard)

## What is DialectOS?
DialectOS is an open-source Spanish dialect translation correctness layer built on the Model
Context Protocol (MCP). It provides 17 translation tools for AI assistants and a CLI
for developers. It supports 25 regional Spanish variants with structure-preserving
translation, glossary enforcement, adversarial quality gates, and standalone validation.

## Key Capabilities
- Translate to 25 Spanish dialects (es-MX, es-AR, es-CO, es-ES, es-PR, etc.)
- Validate existing translations from any provider (`dialectos validate`)
- Preserve markdown structure (tables, code blocks, links)
- i18n locale file diff/merge
- Gender-neutral language support (elles, latine, -e/-x)
- Formality checking (tú vs usted)
- Semantic drift detection
- Translation memory with SHA-256 keyed caching
- Translation corpus — dialect-indexed JSONL storage for translations and corrections
- Auto-glossary — learns terminology corrections from user feedback
- Public benchmark — 205 adversarial samples across 25 dialects and 13 categories
- GitHub Action — CI validation on every pull request
- Automated tests across 7 packages

## Use Cases
- Validating that translations are correct for a specific dialect (standalone, no provider switch needed)
- Localizing SaaS apps for Latin American markets
- Translating technical documentation to regional Spanish
- Maintaining consistent terminology across dialects
- Quality-gating AI-generated translations
- CI validation of locale files on every PR
- MCP-native translation workflows in Claude/Cursor

## CLI Commands
- `dialectos translate` — translate text to a specific dialect
- `dialectos translate-readme` — full README translation with structure preservation
- `dialectos translate-api-docs` — API documentation translation
- `dialectos validate` — validate existing translations against quality checks
- `dialectos corpus stats` — view translation corpus statistics
- `dialectos corpus export` — export corpus to JSONL
- `dialectos benchmark run` — run dialect quality benchmark
- `dialectos benchmark report` — render benchmark results
- `dialectos glossary suggest` — auto-generate glossary from corrections
- `dialectos glossary diff` — compare two glossary versions
- `dialectos i18n detect-missing` — find missing locale keys
- `dialectos i18n batch-translate` — translate to multiple dialects
- `dialectos dialects list` — list all 25 supported dialects
- `dialectos dialects detect` — detect dialect from sample text

## Comparison with Alternatives
| Feature | Google Translate | DeepL | DialectOS |
|---------|----------------|-------|-----------|
| Spanish dialects | 1 (generic) | ~5 | 25 |
| MCP integration | No | No | Yes (17 tools) |
| Markdown preservation | No | No | Yes |
| i18n locale support | No | No | Yes |
| Gender-neutral output | No | No | Yes |
| Quality gates | No | No | Yes |
| Standalone validation | No | No | Yes |
| CI integration | No | No | Yes (GitHub Action) |
| Auto-glossary | No | No | Yes |
| Public benchmark | No | No | Yes (205 adversarial samples) |
| Open-source | No | No | Yes (Apache-2.0) |

## License
Apache-2.0.

## Links
- Repository: https://github.com/KyaniteLabs/DialectOS
- Documentation: https://kyanitelabs.github.io/DialectOS
- Issues: https://github.com/KyaniteLabs/DialectOS/issues
- GitHub Action: https://github.com/KyaniteLabs/DialectOS/blob/main/action.yml