Proposal: document AgentsKit as a portable Supabase pgvector adapter for JavaScript agents

[EmersonBraun]

Problem

JavaScript developers building agents with Supabase often have to assemble three separate pieces themselves: a vector-memory contract, PostgREST mutations, and a pgvector search RPC. The easy shortcut is a generic SQL-execution RPC, but exposing caller-provided SQL—especially through a security definer function—is not a pattern I think the Supabase community should encourage.

There is also a portability problem: if application code is written directly around one vector provider's API, moving between Supabase, local pgvector, or another managed backend becomes a rewrite instead of a configuration change.

What is already implemented

AgentsKit is an open-source, modular agent toolkit for JavaScript. Its @agentskit/memory package now includes a Supabase adapter behind the same VectorMemory contract used by its other storage backends.

The Supabase path intentionally uses:

• direct supabase.from(table).upsert(...) for writes;
• direct delete().in('id', ids) for deletion;
• one purpose-specific match_agentskit_vectors RPC for similarity search;
• security invoker, a fixed return shape, metadata containment filters, a threshold, and a hard maximum of 100 results;
• a server-only service credential and lazy, optional loading of @supabase/supabase-js;
• no RPC that accepts arbitrary SQL.

Source: https://github.com/AgentsKit-io/agentskit/blob/e47f30cf5a938dcf865e9342a41fcf9d7d378fd1/packages/memory/src/vector/supabase.ts

Current setup and security guide: https://github.com/AgentsKit-io/agentskit/blob/e47f30cf5a938dcf865e9342a41fcf9d7d378fd1/apps/docs-next/content/docs/data/memory/supabase-vector.mdx

Validation evidence

I tested the frozen implementation against a disposable Supabase free project in sa-east-1:

• direct upsert succeeded for three synthetic vectors;
• similarity search returned the expected order on the first attempt, with scores 1 and 0.993883748801337;
• topK, threshold, and JSON metadata filtering were honored;
• direct deletion succeeded and a follow-up filtered search returned zero records;
• RLS was enabled, the search function was granted only to service_role, and the project and credentials were deleted after validation.

Proposed contribution

If this is useful to Supabase users, I would like to contribute a concise documentation entry rather than change Supabase core or supabase-js. It would contain:

1. the exact table and bounded search-function SQL;
2. a short server-side AgentsKit example;
3. RLS, credential, and security invoker guidance;
4. cleanup instructions;
5. links to the open-source adapter and its tests.

Would this fit best as an AI integration, a pgvector framework example, or another existing docs section? I am happy to follow the maintainers' preferred scope and location before opening a PR.