[schemas] Knowledge graph tables and extraction trigger

[alanshurafa]

Summary

• Adds 5 tables for a knowledge graph layer: entities, edges, thought_entities, entity_extraction_queue, consolidation_log
• Installs auto-queue trigger on thoughts that enqueues new/updated thoughts for entity extraction
• Part of the OB1 alpha milestone (Wave 2 schema — depends on PR [schemas] Enhanced thoughts columns and utility RPCs #1 enhanced-thoughts)

What's included

Object	Type	Purpose
entities	table	Canonical graph nodes (person, project, topic, tool, organization, place)
edges	table	Typed relationships between entities with support counts and confidence
thought_entities	table	Evidence-bearing links from thoughts to entities
entity_extraction_queue	table	Async queue for pending entity extraction work
consolidation_log	table	Audit trail for dedup merges and quality operations
queue_entity_extraction()	trigger function	Auto-queues thoughts on insert/update with fingerprint guards
trg_queue_entity_extraction	trigger	Attached to thoughts table
8 indexes	indexes	Entity type, normalized name, edge traversal, thought links, pending queue

Design decisions

• Idempotent DDL: All CREATE TABLE IF NOT EXISTS, CREATE OR REPLACE FUNCTION, and DROP TRIGGER IF EXISTS patterns — safe to run multiple times
• Trigger guards: Skips system-generated artifacts (metadata.generated_by IS NOT NULL) and no-op updates where content fingerprint hasn't changed
• Queue dedup: ON CONFLICT (thought_id) DO UPDATE only resets when source_fingerprint IS DISTINCT FROM the stored value, preventing churn from metadata-only writes
• Cascade cleanup: Deleting a thought or entity cascades to remove related links and queue entries
• Optional backfill: Commented SQL at bottom for existing brains to queue pre-existing thoughts — not auto-run to avoid surprise load on large brains
• Read access: anon/authenticated get SELECT on all graph tables for MCP tools and REST API queries

Dependencies

• Requires enhanced-thoughts schema (PR [schemas] Enhanced thoughts columns and utility RPCs #1) for enriched thoughts table columns
• Requires content_fingerprint column on thoughts (from Step 2.6 of getting-started guide)
• No external services or API keys needed

Test plan

• Paste schema.sql into Supabase SQL Editor and run — no errors
• Confirm 5 new tables appear in Table Editor
• Confirm queue_entity_extraction function in Database > Functions
• Confirm trg_queue_entity_extraction trigger on thoughts table
• Insert a thought and verify a row appears in entity_extraction_queue
• Re-run the migration — no errors (idempotency check)

🤖 Generated with Claude Code

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 087c43f333

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Use UUID thought_id to match thoughts.id

The new schema defines thought_id as BIGINT, but the documented base Open Brain schema creates thoughts.id as UUID (docs/01-getting-started.md, lines 67-74). In Postgres, this makes the FK invalid (BIGINT cannot reference UUID), so applying this migration will fail when creating thought_entities (and similarly entity_extraction_queue), and the trigger path that inserts NEW.id into the queue will also be type-incompatible. This blocks installation for standard deployments that followed the setup guide.

Useful? React with 👍 / 👎.