rustvani 0.2.2

Voice AI framework for Rust — real-time speech pipelines with STT, LLM, TTS, and Dhara conversation flows
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54

# Neon Postgres Tool

**Files:** `src/tools/postgres/`  
**Feature:** `db-postgres` (not enabled by default)

Built-in LLM tool for Neon Postgres with schema caching, parameterized queries, pgvector similarity search, and structured filters. The LLM never writes raw SQL — all queries are parameterized.

## Setup

```toml
[dependencies]
rustvani = { version = "0.2", features = ["db-postgres"] }
```

```bash
DATABASE_URL=postgres://user:pass@host/db
```

## Usage

```rust
use rustvani::tools::postgres::NeonPostgresTool;
use rustvani::services::llm::openai::OpenAILLMHandler;

let pg = Arc::new(NeonPostgresTool::from_env());

let mut llm = OpenAILLMHandler::new(config);
llm.add_tool(pg);
let processor = llm.into_processor();
```

## Registered Functions

When attached to an `OpenAILLMHandler`, the tool registers these functions into the shared `FunctionRegistry`:

| Function | Purpose |
|---|---|
| `pg_schema` | Introspect tables, columns, and types |
| `pg_query` | Execute a parameterized SELECT |
| `pg_refine` | Refine a query plan based on schema |
| `pg_vector_search` | pgvector similarity search with structured filters |

## Lifecycle

- `on_start` (cacheable): Connects to Postgres and caches schema
- `on_stop`: Returns connections, flushes caches
- `on_cancel`: Cancels in-flight queries, then `on_stop`

## Cargo Feature

```toml
[dependencies]
rustvani = { version = "0.2", features = ["db-postgres"] }
```