Add embedding columns for tasks and tickets

[dolliecoder]

This PR adds embedding columns for task and ticket descriptions as a schema-only, foundational step toward RAG support.

What changed :

Added a nullable description_embedding vector column to:
tasks
tickets
Vector dimensions match the existing meeting embeddings for consistency.

Before :

Tasks and tickets had no schema support for storing embeddings.
Future semantic search or RAG work for these entities was blocked at the database layer.

After :

Tasks and tickets can now store vector embeddings.
No existing behavior, queries, or APIs are affected.

Why this is safe:

Schema-only change (no triggers, no generation logic).
Nullable columns with no runtime usage.
No behavioral or AI-pipeline changes.
Fully reversible.

Context :

This PR is a small, incremental step toward Issue #65.
Follow-up PRs will handle vector search and embedding generation separately to keep reviews focused and low-risk.

Summary by CodeRabbit

• Chores
  • Added embedding vector columns to tasks and tickets to support advanced search and semantic capabilities.
  • Enabled vector support in the database to allow storage and querying of embeddings.

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds Postgres vector extension and adds description_embedding vector(768) columns to the tasks and tickets tables to store 768-dimensional description embeddings.

Changes

Cohort / File(s)	Summary
DB Extensions sqls/01_user_auth_schema.sql	Adds CREATE EXTENSION IF NOT EXISTS vector; (keeps existing uuid-ossp extension).
Vector Embedding Columns sqls/03_task_schema.sql, sqls/04_tickets_schema.sql	Added description_embedding vector(768) column to tasks and tickets tables respectively.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~5 minutes

Poem

🐇 I nibble at schemas in moonlit night,
New vectors tucked in rows so tight,
Seven-sixty-eight whispers, soft and keen,
Tasks and tickets dreaming in-between,
Hop, hum, embed — 🎶 a rabbit's delight.

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately summarizes the main change: adding embedding columns to tasks and tickets tables, which is reflected in all three modified schema files.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

No actionable comments were generated in the recent review. 🎉

🧹 Recent nitpick comments

sqls/01_user_auth_schema.sql (1)

1-4: Consider adding an inline comment for the vector extension.

Line 1's comment only describes uuid-ossp. A brief comment for the vector extension (e.g., its purpose for embedding columns) would keep things consistent and help future readers understand why it's here.

 -- Installs uuid-ossp extension for UUID generation
 CREATE EXTENSION IF NOT EXISTS "uuid-ossp";
+-- Installs pgvector extension for vector/embedding column support
 CREATE EXTENSION IF NOT EXISTS vector;

Otherwise, the change is correct — IF NOT EXISTS makes it idempotent, and placing it in this file ensures the type is available before the downstream vector(768) columns in 03_task_schema.sql and 04_tickets_schema.sql.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@sqls/04_tickets_schema.sql`:
- Line 7: The tickets schema migration uses the pgvector type via the
description_embedding column (vector(768)) before the extension is created; add
a CREATE EXTENSION IF NOT EXISTS vector; statement to an earlier migration or
directly at the top of 04_tickets_schema.sql (before any CREATE TABLE that
references vector) so the vector type exists when description_embedding is
declared, and remove or keep idempotent the later CREATE EXTENSION in
09_meeting_vector_search.sql to avoid ordering failures.

[dolliecoder]

moved pgvector extension creation to an early shared schema so vector types are available before task/ticket tables are defined