facebook/duckling

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

4313 stars · 740 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Duckling is a Haskell library that parses natural language text into structured data, extracting entities like dates, times, amounts of money, numbers, and durations. It ships with an example HTTP server so non-Haskell backends can use it via a simple API.

Use cases

• parse dates and times from user text like 'tomorrow at eight'
• extract amounts of money and durations from chat messages
• recognize credit card numbers and phone numbers in input strings
• build a chatbot that understands quantities and ordinals
• run an entity-extraction HTTP service for a non-Haskell backend

When to choose

• you need deterministic, rule-based entity extraction without training data
• you're building a conversational assistant that must understand dates, times, money, and numbers
• you want a self-hosted alternative to wit.ai's built-in entities

When to avoid

• you need general-purpose NLP like sentiment analysis or named-entity recognition of people and places
• your stack can't run Haskell or host a separate parsing service
• you need ML-based extraction that adapts to novel phrasings

Facets

library · maturity active

nlp parser http-server developer-tools windows entity-extraction haskell text-parsing date-parsing wit-ai natural-language-processing linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebook/duckling")