vector-store-wasp-drone — Claude Code subagent: pgvector, Drizzle
vector-store-wasp-drone owns embedding storage: column shape, index choice, migrations, dataset versions. Useful when a project has to settle on pgvector, Deep Lake or Qdrant.
AGPL-3.0-or-later ★ 96 updated 2026-10-18
/plugin marketplace add legioncodeinc/vibe-coding-tools
/plugin install data-and-ai-tools@wasp-nestWhat vector-store-wasp-drone does
The subagent covers embedding storage: column shape, index choice, migrations and dataset versioning. Neon Postgres with pgvector and Drizzle is the stack it assumes first. Deep Lake, Qdrant and managed services are documented as alternatives, with a matrix for picking between them.
Schema work goes through Drizzle's vector({ dimensions: N }), sized to whatever the embedding model outputs. HNSW is the default index, with the operator class matched to the distance the queries actually use. A vector_l2_ops index does nothing for a <=> query — no error, just a sequential scan, so confirm with EXPLAIN before deciding the index is broken.
Hybrid search pairs a vector column with a generated tsvector column and a GIN index; fusion and ranking belong to another subagent. Changing dimensions is a four-step migration: new column, backfill, cutover, cleanup. On Claude the agent file sits at .claude/agents/vector-store-wasp-drone.md; for Codex it is .codex/agents/vector-store-wasp-drone.toml.
What it helps with
- Design a vector column that matches the model's output width
- Pick between HNSW and IVFFlat for a given query
- Wire hybrid search over
vectorandtsvectorcolumns - Plan a dimension change as a migration instead of a resize
- Decide whether pgvector is enough or Qdrant is warranted
Example prompts for the agent
- Design an embedding table in Drizzle sized to the model's output width
- Which index should this vector column use if we search by cosine distance?
- Wire hybrid search over the
vectorandtsvectorcolumns here
How to install vector-store-wasp-drone
1Plugin marketplace
/plugin marketplace add legioncodeinc/vibe-coding-tools/plugin install data-and-ai-tools@wasp-nest2Manually: put the files in place
.claude/agents/vector-store-wasp-drone.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.codex/agents/vector-store-wasp-drone.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/vector-store-wasp-drone.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about vector-store-wasp-drone
How to install vector-store-wasp-drone in Claude Code?
Run /plugin marketplace add legioncodeinc/vibe-coding-tools, then /plugin install data-and-ai-tools@wasp-nest. The agent file lands at .claude/agents/vector-store-wasp-drone.md.
What does vector-store-wasp-drone do?
It designs vector storage: the column and its dimensions, the index with a matching operator class, hybrid search wiring, migrations and dataset versions. The default stack is Neon Postgres with pgvector and Drizzle.
What does it hand off to other subagents?
Schema PRD writing, secret and PII audits, and recall tuning or embedding-model choice stay with other wasp-drones. This one flags those concerns and passes them on.