feat: add query/retrieval FlowProcessor services and missing runner scripts

Wire up the query and retrieval side of the pipeline so the agent can
answer questions from stored knowledge:

- Triples query service (FalkorDB) — all SPO pattern queries via NATS
- Graph embeddings query service (Qdrant) — entity vector similarity
- Document embeddings query service (Qdrant) — chunk vector similarity
- Graph RAG service — full concept→entity→traverse→score→synthesize pipeline
- Document RAG service — embed→find chunks→synthesize pipeline
- Runner scripts for chunker, extractor, embeddings (missing from Phase 5)
- Add DocumentEmbeddingsRequest/Response schema types
- Add RAG prompt templates (extract-concepts, edge-scoring, synthesize)
- Add graph/doc embeddings query topics to seed config + flow manager
- Add all pipeline/query/retrieval services to docker-compose
- 8 new runner scripts, 8 new pnpm script aliases

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

ts/packages/flow/src/query/embeddings/qdrant-doc-service.ts

@@ -0,0 +1,81 @@
+/**
+ * Document embeddings query service — finds similar document chunks in Qdrant.
+ *
+ * Wraps QdrantDocEmbeddingsQuery as a NATS consumer so Document RAG can look up
+ * chunks by vector similarity over the message bus.
+ *
+ * Python reference: trustgraph-flow/trustgraph/query/doc_embeddings/qdrant/service.py
+ */
+
+import {
+  FlowProcessor,
+  ConsumerSpec,
+  ProducerSpec,
+  type ProcessorConfig,
+  type FlowContext,
+  type DocumentEmbeddingsRequest,
+  type DocumentEmbeddingsResponse,
+} from "@trustgraph/base";
+import { QdrantDocEmbeddingsQuery } from "./qdrant-doc.js";
+
+export class DocEmbeddingsQueryService extends FlowProcessor {
+  private query: QdrantDocEmbeddingsQuery;
+
+  constructor(config: ProcessorConfig) {
+    super(config);
+    this.query = new QdrantDocEmbeddingsQuery();
+
+    this.registerSpecification(
+      new ConsumerSpec<DocumentEmbeddingsRequest>(
+        "document-embeddings-request",
+        this.onMessage.bind(this),
+      ),
+    );
+    this.registerSpecification(
+      new ProducerSpec<DocumentEmbeddingsResponse>("document-embeddings-response"),
+    );
+
+    console.log("[DocEmbeddingsQuery] Service initialized");
+  }
+
+  private async onMessage(
+    msg: DocumentEmbeddingsRequest,
+    properties: Record<string, string>,
+    flowCtx: FlowContext,
+  ): Promise<void> {
+    const requestId = properties.id;
+    if (!requestId) return;
+
+    const producer = flowCtx.flow.producer<DocumentEmbeddingsResponse>("document-embeddings-response");
+    const collection = msg.collection ?? "default";
+
+    try {
+      const allChunks: DocumentEmbeddingsResponse["chunks"] = [];
+
+      for (const vector of msg.vectors ?? []) {
+        const matches = await this.query.query({
+          vector,
+          user: "default",
+          collection,
+          limit: msg.limit ?? 10,
+        });
+
+        for (const match of matches) {
+          allChunks.push({ chunkId: match.chunkId, score: match.score });
+        }
+      }
+
+      await producer.send(requestId, { chunks: allChunks });
+    } catch (err) {
+      console.error("[DocEmbeddingsQuery] Query failed:", err);
+      await producer.send(requestId, {
+        chunks: [],
+        error: { type: "query-error", message: String(err) },
+      });
+    }
+  }
+}
+
+export async function run(): Promise<void> {
+  await DocEmbeddingsQueryService.launch("doc-embeddings-query");
+}

ts/packages/flow/src/query/embeddings/qdrant-graph-service.ts

@@ -0,0 +1,83 @@
+/**
+ * Graph embeddings query service — finds similar entities in Qdrant via FlowProcessor.
+ *
+ * Wraps QdrantGraphEmbeddingsQuery as a NATS consumer so Graph RAG can look up
+ * entities by vector similarity over the message bus.
+ *
+ * Python reference: trustgraph-flow/trustgraph/query/graph_embeddings/qdrant/service.py
+ */
+
+import {
+  FlowProcessor,
+  ConsumerSpec,
+  ProducerSpec,
+  type ProcessorConfig,
+  type FlowContext,
+  type GraphEmbeddingsRequest,
+  type GraphEmbeddingsResponse,
+} from "@trustgraph/base";
+import { QdrantGraphEmbeddingsQuery } from "./qdrant-graph.js";
+
+export class GraphEmbeddingsQueryService extends FlowProcessor {
+  private query: QdrantGraphEmbeddingsQuery;
+
+  constructor(config: ProcessorConfig) {
+    super(config);
+    this.query = new QdrantGraphEmbeddingsQuery();
+
+    this.registerSpecification(
+      new ConsumerSpec<GraphEmbeddingsRequest>(
+        "graph-embeddings-request",
+        this.onMessage.bind(this),
+      ),
+    );
+    this.registerSpecification(
+      new ProducerSpec<GraphEmbeddingsResponse>("graph-embeddings-response"),
+    );
+
+    console.log("[GraphEmbeddingsQuery] Service initialized");
+  }
+
+  private async onMessage(
+    msg: GraphEmbeddingsRequest,
+    properties: Record<string, string>,
+    flowCtx: FlowContext,
+  ): Promise<void> {
+    const requestId = properties.id;
+    if (!requestId) return;
+
+    const producer = flowCtx.flow.producer<GraphEmbeddingsResponse>("graph-embeddings-response");
+    const user = msg.collection ?? "default";
+    const collection = msg.collection ?? "default";
+
+    try {
+      // Query for each vector and aggregate results
+      const allEntities: GraphEmbeddingsResponse["entities"] = [];
+
+      for (const vector of msg.vectors ?? []) {
+        const matches = await this.query.query({
+          vector,
+          user,
+          collection,
+          limit: msg.limit ?? 50,
+        });
+
+        for (const match of matches) {
+          allEntities.push(match.entity);
+        }
+      }
+
+      await producer.send(requestId, { entities: allEntities });
+    } catch (err) {
+      console.error("[GraphEmbeddingsQuery] Query failed:", err);
+      await producer.send(requestId, {
+        entities: [],
+        error: { type: "query-error", message: String(err) },
+      });
+    }
+  }
+}
+
+export async function run(): Promise<void> {
+  await GraphEmbeddingsQueryService.launch("graph-embeddings-query");
+}