vector type in RPC function parameters generates incorrect TypeScript types

[RasmusKD]

Bug report

Describe the bug

When using supabase gen types typescript with RPC functions that have vector parameters, the generated types are string, but the actual Supabase client requires number[] to work correctly.
This is different from the closed issue #578 which focused on table columns. For RPC functions, the workaround of using JSON.stringify() does NOT work - it causes the query to fail or return zero results.

To Reproduce

1. Create a hybrid search function using pgvector:

CREATE OR REPLACE FUNCTION match_documents(
  query_embedding vector(1536),
  match_count int DEFAULT 5
)
RETURNS TABLE (
  id uuid,
  content text,
  similarity float
)
LANGUAGE plpgsql
AS $$
BEGIN
  RETURN QUERY
  SELECT
    documents.id,
    documents.content,
    (1 - (documents.embedding <=> query_embedding)) AS similarity
  FROM documents
  ORDER BY documents.embedding <=> query_embedding
  LIMIT match_count;
END;
$$;

2. Generate types:
   supabase gen types typescript --project-id <project-id> > types/supabase.ts

3. Observe generated types:

Functions: {
  match_documents: {
    Args: {
      query_embedding: string  // ❌ Generated as string
      match_count?: number
    }
    Returns: { /* ... */ }[]
  }
}

4. Try to call the function:

// Get embedding from OpenAI (returns number[])
const embedding = [0.1, 0.2, 0.3, /* ... 1536 floats */];

// ❌ TypeScript error: number[] is not assignable to string
const { data } = await supabase.rpc("match_documents", {
  query_embedding: embedding,
});

Expected behavior
The generated type for vector parameters should be number[]:

Functions: {
  match_documents: {
    Args: {
      query_embedding: number[]  // ✅ Should be number[]
      match_count?: number
    }
  }
}

What DOES work vs what DOESN'T:

INSERT into table column:

• JSON.stringify() → ✅ Works
• Raw number[] → ❌ Fails

RPC function parameter:

• JSON.stringify() → ❌ Fails
• Raw number[] → ✅ Works

This inconsistency makes it impossible to have correct types for both use cases.
Current workarounds (all bad)

Use any type for Supabase client - loses all type safety
Use @ts-expect-error - suppresses errors but doesn't fix the type
Manually edit generated types - gets overwritten on regeneration

Environment

Supabase CLI version: latest
pgvector extension: enabled
TypeScript version: 5.x
supabase-js version: 2.86.2

Why this matters

Vector similarity search via RPC functions is a common pattern for AI/RAG applications. The current type generation makes it impossible to build type-safe applications using this pattern.

Proposed solution
Map vector and vector(n) types to number[] in TypeScript generation for function parameters.

Related issues

#578 - Closed as not planned, but only addressed table columns, not RPC functions