cqs 1.26.0

Code intelligence and RAG for AI agents. Semantic search, call graphs, impact analysis, type dependencies, and smart context assembly — in single tool calls. 54 languages + L5X/L5K PLC exports, 91.2% Recall@1 (BGE-large), 0.951 MRR (296 queries). Local ML, GPU-accelerated.
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

//! Central scoring configuration constants.

/// Central configuration for all search scoring constants.
/// Consolidates name matching tiers, note boost factor, importance
/// demotion weights, and parent boost parameters into one struct.
/// Use `ScoringConfig::DEFAULT` everywhere — no scattered magic numbers.
pub(crate) struct ScoringConfig {
    pub name_exact: f32,
    pub name_contains: f32,
    pub name_contained_by: f32,
    pub name_max_overlap: f32,
    pub note_boost_factor: f32,
    pub importance_test: f32,
    pub importance_private: f32,
    pub parent_boost_per_child: f32,
    pub parent_boost_cap: f32,
}

impl ScoringConfig {
    pub const DEFAULT: Self = Self {
        name_exact: 1.0,
        name_contains: 0.8,
        name_contained_by: 0.6,
        name_max_overlap: 0.5,
        note_boost_factor: 0.15,
        importance_test: 0.70,
        importance_private: 0.80,
        parent_boost_per_child: 0.05,
        parent_boost_cap: 1.15,
    };

    // CQ-7: `with_overrides` removed — was dead code (`#[allow(dead_code)]`).
    // Scoring overrides from `[scoring]` in config are applied per-field where needed
    // (e.g., `rrf_k` via `set_rrf_k_from_config`). If full config-driven scoring is
    // added later, re-introduce this method with actual production callers.
}