feat: Phase 2 — embeddings + hybrid memory.search

Adds a small embedder sidecar (Xenova/bge-small-en-v1.5, ONNX, CPU-only)
that the web app calls inline on memory.write and memory.update, and on
demand from the new memory.search tool.

memory.search performs three candidate fetches in parallel — pgvector
cosine similarity, Postgres full-text via plainto_tsquery + ts_rank_cd,
and tag-set overlap — then fuses them with Reciprocal Rank Fusion
(k=60). Each result carries its per-source rank so the model can see
*why* a memory surfaced.

The migrator boot step gained an idempotent embedding backfill: any row
with embedding IS NULL is batched (32 at a time) through the embedder
after SQL migrations apply. Safe to run on every boot.

New tool memory.update fixes the missing edit path; centralises the
re-embed-on-content-change rule alongside write.

Stack additions:
- apps/embedder/ — Fastify server, persistent /data/models volume so the
  ~30 MB model only downloads once
- apps/web/lib/embedder.ts — typed HTTP client with batched embed +
  health probe
- packages/schemas — MemoryUpdateInput, MemorySearchInput
- docker-compose — embedder service, healthcheck, app + migrator both
  depend_on it healthy; EMBEDDER_URL promoted to a required env var

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-05-15 09:04:44 -07:00
co-authored by Claude Opus 4.7
parent e1fa1197f4
commit 9a8b504f51
12 changed files with 1523 additions and 19 deletions
+18
View File
@@ -42,6 +42,24 @@ export const MemoryIdInput = z.object({
});
export type MemoryIdInput = z.infer<typeof MemoryIdInput>;
export const MemoryUpdateInput = z.object({
id: z.string().uuid(),
content: MemoryContent.optional(),
tags: Tags.optional(),
}).refine((v) => v.content !== undefined || v.tags !== undefined, {
message: "memory.update requires content or tags",
});
export type MemoryUpdateInput = z.infer<typeof MemoryUpdateInput>;
export const MemorySearchInput = z.object({
query: z.string().min(1).max(2000),
project: ProjectKey.optional(),
scope: MemoryScope.optional(),
tags: z.array(z.string()).optional(),
limit: z.number().int().min(1).max(50).default(10),
});
export type MemorySearchInput = z.infer<typeof MemorySearchInput>;
export const ProjectIdentifyInput = z.object({
key: ProjectKey,
display_name: z.string().min(1).max(200).optional(),