vtcode-core 0.99.2

Core library for VT Code - a Rust-based terminal coding agent
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
135
136

use std::hint::black_box;
use std::sync::Arc;

use criterion::Criterion;
use criterion::criterion_group;
use criterion::criterion_main;
use tokio::runtime::Runtime;
use tokio::sync::RwLock;
use vtcode_core::core::agent::harness_kernel::{
    HarnessRequestPlanInput, PreparedToolBatch, PreparedToolCall, build_harness_request_plan,
};
use vtcode_core::llm::provider::{Message, ToolChoice, ToolDefinition};
use vtcode_core::tools::registry::SessionToolCatalogState;

fn sample_tool(name: &str) -> ToolDefinition {
    ToolDefinition::function(
        name.to_string(),
        format!("Tool {name}"),
        serde_json::json!({
            "type": "object",
            "properties": {
                "path": { "type": "string" }
            }
        }),
    )
}

fn sample_tools(count: usize) -> Arc<Vec<ToolDefinition>> {
    Arc::new(
        (0..count)
            .map(|index| sample_tool(&format!("tool_{index}")))
            .collect(),
    )
}

fn sample_messages(count: usize) -> Vec<Message> {
    (0..count)
        .map(|index| Message::user(format!("message {index}")))
        .collect()
}

fn request_plan_benchmark(c: &mut Criterion) {
    let tools = sample_tools(24);
    let messages = sample_messages(32);

    c.bench_function("agent_harness_request_plan_with_tools", |b| {
        b.iter(|| {
            black_box(build_harness_request_plan(HarnessRequestPlanInput {
                messages: messages.clone(),
                system_prompt: "System prompt\n[Runtime Context]\nturn=12".to_string(),
                tools: Some(Arc::clone(&tools)),
                model: "gpt-5".to_string(),
                max_tokens: Some(2000),
                temperature: Some(0.7),
                stream: true,
                tool_choice: Some(ToolChoice::auto()),
                parallel_tool_config: None,
                reasoning_effort: None,
                verbosity: None,
                metadata: None,
                context_management: None,
                previous_response_id: Some("resp_123".to_string()),
                prompt_cache_key: Some("session:test".to_string()),
                prompt_cache_profile: None,
                tool_catalog_hash: None,
            }))
        })
    });
}

fn prepared_batch_planning_benchmark(c: &mut Criterion) {
    let calls: Vec<PreparedToolCall> = (0..48)
        .map(|index| {
            let readonly = index % 5 != 0;
            PreparedToolCall::new(
                format!("tool_{index}"),
                readonly,
                readonly,
                serde_json::json!({ "path": format!("src/file_{index}.rs") }),
            )
        })
        .collect();

    c.bench_function("agent_harness_prepared_batch_plan", |b| {
        b.iter(|| black_box(PreparedToolBatch::plan(calls.clone(), true)))
    });
}

fn tool_catalog_projection_benchmark(c: &mut Criterion) {
    let runtime = Runtime::new().expect("criterion tokio runtime");
    let state = Arc::new(SessionToolCatalogState::new());
    let tools = Arc::new(RwLock::new((*sample_tools(32)).clone()));

    runtime.block_on(async {
        let _ = state
            .filtered_snapshot_with_stats(&tools, true, false)
            .await;
    });

    c.bench_function("agent_harness_tool_catalog_cache_hit", |b| {
        b.iter(|| {
            let state = Arc::clone(&state);
            let tools = Arc::clone(&tools);
            runtime.block_on(async move {
                black_box(
                    state
                        .filtered_snapshot_with_stats(&tools, true, false)
                        .await,
                )
            })
        })
    });

    c.bench_function("agent_harness_tool_catalog_cache_miss", |b| {
        b.iter(|| {
            let state = Arc::clone(&state);
            let tools = Arc::clone(&tools);
            runtime.block_on(async move {
                state.note_explicit_refresh("benchmark");
                black_box(
                    state
                        .filtered_snapshot_with_stats(&tools, true, false)
                        .await,
                )
            })
        })
    });
}

criterion_group!(
    benches,
    request_plan_benchmark,
    prepared_batch_planning_benchmark,
    tool_catalog_projection_benchmark
);
criterion_main!(benches);