Example iOS app that seeds a task management SQLite database and
renders SwiftDBAI's DataChatView. Ships with a mock LLM for offline
use and an OllamaWithSystemPrompt wrapper that fixes AnyLanguageModel's
Ollama adapter dropping system instructions.
Tested successfully with:
- DemoLanguageModel (mock, pattern-matched SQL)
- Ollama qwen3-coder-30b (local, real SQL generation)
- OpenAI gpt-4o-mini (JOINs, GROUP BY, multi-turn)
- Anthropic claude-haiku-4.5 (complex aggregations)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Chat with any SQLite database using natural language. Built on
AnyLanguageModel (HuggingFace) for LLM-agnostic provider support
and GRDB for SQLite access.
Core features:
- Auto schema introspection from sqlite_master (zero config)
- NL → SQL generation via any AnyLanguageModel provider
- Three rendering modes: text summary, data table, Swift Charts
- Drop-in DataChatView (SwiftUI) and headless ChatEngine
- Operation allowlist with read-only default
- Mutation policy with per-table control
- ToolExecutionDelegate for destructive operation confirmation
- Multi-turn conversation context
- 352 tests across 24 suites, all passing
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
