Context Pipeline

Implementation status (2026-07-23, kelly): ✅ Implemented. src/context/mod.rs — ContextPipeline/ContextPipelineConfig (defaults max_entities:20, expand_links:true, link_depth:1), KnowledgeContext/KnowledgeEntity, the Direct/Linked/Semantic relevance enum (Semantic from the hybrid vector path), and query(). Exposed as the quipu_context MCP tool and the /context REST route. Verified by grep.

The context pipeline blends knowledge graph facts with code context, producing unified results for agent consumption. It’s the integration surface between Quipu and Bobbin.

How It Works

When an agent asks for context about a topic:

1. Text search – SPARQL FILTER(CONTAINS(...)) on entity IRIs and literal values to find direct hits
2. Link expansion – follow outgoing and incoming relationships from direct hits to discover related entities
3. Rank and truncate – sort by relevance score, trim to budget

The output is a KnowledgeContext shaped for Bobbin to merge with its code search results.

Output Shape

KnowledgeContext

{
  "query": "traefik",
  "entities": [ ... ],
  "summary": {
    "total_entities": 4,
    "total_facts": 18,
    "direct_hits": 1,
    "linked_additions": 3
  }
}

KnowledgeEntity

Each entity includes its label, types, relevance, and all its facts:

{
  "iri": "http://example.org/traefik",
  "label": "Traefik",
  "types": ["http://example.org/WebApplication"],
  "relevance": "Direct",
  "score": 1.0,
  "facts": [
    { "predicate": "http://example.org/runsOn", "value": "http://example.org/kota", "value_type": "Entity" },
    { "predicate": "http://example.org/port", "value": "443", "value_type": "Literal" }
  ]
}

Relevance Types

Configuration

MCP Tool

{
  "tool": "quipu_context",
  "input": {
    "query": "traefik reverse proxy",
    "max_entities": 10,
    "expand_links": true
  }
}

REST API

curl -s localhost:3030/context -X POST \
  -H "Content-Type: application/json" \
  -d '{"query": "traefik", "max_entities": 10}'

Rust API

#![allow(unused)]
fn main() {
use quipu::context::{ContextPipeline, ContextPipelineConfig};

let config = ContextPipelineConfig {
    max_entities: 10,
    expand_links: true,
    ..Default::default()
};

let pipeline = ContextPipeline::new(&store, config);
let ctx = pipeline.query("traefik").unwrap();

for entity in &ctx.entities {
    println!("{} ({:?}): {} facts",
        entity.label.as_deref().unwrap_or(&entity.iri),
        entity.relevance,
        entity.facts.len());
}
}