ITCy has no migration tree. Schema and statements are the sql/*.sql files in this repository. The running process opens SQLite and applies those statements.
sql/runtime.db (plus -wal / -shm while open) is created on first open. It holds sources, chunks, Slack memory, drafts, digest, and publish audit. Do not commit *.db files.
A separate scrape cache may exist as sql/linkedin-scrape-cache.db (also gitignored).
| File | Role |
|---|---|
sql/sources_schema.sql |
Corpus rows (export, enrich, ingest) |
sql/chunks_schema.sql |
RAG chunks |
sql/messages_schema.sql |
Slack thread memory |
sql/drafts_schema.sql |
LinkedIn drafts and X tweets |
sql/pending_drafts_schema.sql |
In-flight draft build |
sql/digest_schema.sql |
Daily digest |
sql/publish_audit_schema.sql |
Ship audit |
sql/scrape_pages_schema.sql |
LinkedIn scrape page cache |
Neighbor *_insert.sql / *_get.sql / *_list.sql files are the statements the Rust store uses. Edit schema and statements together; open the database with the product binary so the file matches the code.