openinference-semantic-conventions 0.1.1

OpenInference semantic conventions for LLM observability in Rust
Documentation 
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
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134

# openinference-rs

Rust implementation of [OpenInference](https://github.com/Arize-ai/openinference) semantic conventions for LLM observability.

[![Crates.io](https://img.shields.io/crates/v/openinference-semantic-conventions.svg)](https://crates.io/crates/openinference-semantic-conventions)
[![Documentation](https://docs.rs/openinference-semantic-conventions/badge.svg)](https://docs.rs/openinference-semantic-conventions)
[![License](https://img.shields.io/crates/l/openinference-semantic-conventions.svg)](LICENSE-MIT)

## Overview

OpenInference is a set of conventions for instrumenting LLM applications, compatible with [OpenTelemetry](https://opentelemetry.io/) and designed to work with observability platforms like [Arize Phoenix](https://phoenix.arize.com/).

This repository provides Rust crates for:

- **Semantic conventions** - Attribute constants matching the [OpenInference specification](https://github.com/Arize-ai/openinference/blob/main/spec/semantic_conventions.md)
- **OTel GenAI compatibility** - Aliases for [OpenTelemetry GenAI semantic conventions](https://opentelemetry.io/docs/specs/semconv/gen-ai/)
- **Instrumentation helpers** - Span builders for easy integration with the `tracing` crate

## Crates

| Crate | Description |
|-------|-------------|
| [`openinference-semantic-conventions`](./openinference-semantic-conventions) | Attribute constants and types |
| [`openinference-instrumentation`](./openinference-instrumentation) | Span builders and helpers |

## Quick Start

Add to your `Cargo.toml`:

```toml
[dependencies]
openinference-semantic-conventions = "0.1"
openinference-instrumentation = "0.1"
```

### Basic Usage

```rust
use openinference_semantic_conventions::{attributes, SpanKind};
use openinference_instrumentation::LlmSpanBuilder;

// Create an LLM span with OpenInference attributes
let span = LlmSpanBuilder::new("gpt-4")
    .provider("openai")
    .temperature(0.7)
    .max_tokens(1000)
    .build();

let _guard = span.enter();
// ... perform LLM call ...
```

### Using Attribute Constants Directly

```rust
use openinference_semantic_conventions::{attributes, gen_ai, SpanKind};
use opentelemetry::KeyValue;

// OpenInference attributes
let kind = KeyValue::new(attributes::OPENINFERENCE_SPAN_KIND, SpanKind::Llm.as_str());
let model = KeyValue::new(attributes::llm::MODEL_NAME, "gpt-4");
let tokens = KeyValue::new(attributes::llm::token_count::TOTAL, 150i64);

// OTel GenAI attributes (for dual compatibility)
let gen_ai_model = KeyValue::new(gen_ai::request::MODEL, "gpt-4");
let input_tokens = KeyValue::new(gen_ai::usage::INPUT_TOKENS, 100i64);
```

## Span Kinds

OpenInference defines the following span kinds:

| Kind | Description |
|------|-------------|
| `LLM` | Call to a Large Language Model |
| `EMBEDDING` | Call to generate embeddings |
| `CHAIN` | Workflow/pipeline step or glue code |
| `TOOL` | External tool/function execution |
| `AGENT` | Reasoning block using LLMs and tools |
| `RETRIEVER` | Vector store or database query |
| `RERANKER` | Document reranking |
| `GUARDRAIL` | Input/output validation |
| `EVALUATOR` | Model output evaluation |

## Dual Attribute Support

This library supports both OpenInference (`llm.*`) and OTel GenAI (`gen_ai.*`) attribute conventions for maximum compatibility:

```rust
use openinference_semantic_conventions::gen_ai;

// Map between conventions
let otel_key = gen_ai::map_openinference_to_gen_ai("llm.model_name");
let oi_key = gen_ai::map_gen_ai_to_openinference("gen_ai.request.model");
```

## Integration with Observability Backends

Spans created with this library are compatible with:

- [Arize Phoenix](https://phoenix.arize.com/) (native OpenInference support)
- [Grafana Tempo](https://grafana.com/oss/tempo/) (via OTel)
- [Jaeger](https://www.jaegertracing.io/) (via OTel)
- [Datadog](https://www.datadoghq.com/) (OTel GenAI support)
- [Honeycomb](https://www.honeycomb.io/) (via OTel)
- Any OpenTelemetry-compatible backend

## Configuration

### Environment Variables

```bash
# Privacy settings (opt-in content recording)
OPENINFERENCE_HIDE_INPUTS=false
OPENINFERENCE_HIDE_OUTPUTS=false

# OTel exporter configuration
OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4317
OTEL_SERVICE_NAME=my-llm-app
```

## License

MIT OR Apache-2.0

## Contributing

Contributions are welcome! Please feel free to submit issues and pull requests.

## Related Projects

- [OpenInference](https://github.com/Arize-ai/openinference) - Original specification and Python/JS implementations
- [Arize Phoenix](https://github.com/Arize-ai/phoenix) - AI observability platform
- [OpenTelemetry Rust](https://github.com/open-telemetry/opentelemetry-rust) - Rust OTel SDK