SQLite-to-Oracle Planner

Do NOT load this skill when

Do not load this skill for unrelated general programming, non-Oracle cloud work, or questions covered by a narrower sibling skill. When the request is only asking to find or install skills, use find-skills instead.

When to Use

Load this skill for: the user asks to "migrate SQLite to Oracle", "replace better-sqlite3", "plan Oracle migration", "convert SQLite schema", or "find SQLite touch points".

Prefer this skill only for its named domain. For broader OCI architecture triage, start with oci/best-practices as the router.

Role

This is a planning-only skill. Search, read minimally, and produce the manifest. Do NOT implement changes. Use Glob and Grep — never read entire files to scan them.

Scan Strategy

Work in five passes. Log every hit with file path and line number.

Pass 1 — Dependency Imports

Pattern: better-sqlite3 | sqlite3 | drizzle-orm/sqlite-core | typeorm.*sqlite | @agent-state
Files:   **/*.ts, **/*.js, **/*.mjs, **/*.cjs

Import Found	Oracle Equivalent
`better-sqlite3`	`oracledb` or project Oracle connection pool
`sqlite3`	`oracledb`
`drizzle-orm/sqlite-core`	Raw OCI SQL or `drizzle-orm` with Oracle adapter
`typeorm` + sqlite driver	`typeorm` with Oracle driver
`@agent-state` (SQLite-backed)	Oracle-backed state store

Pass 2 — SQLite Data Files

Glob: **/*.db, **/*.sqlite, **/*.sqlite3

These are data files — flag each one. They require data migration, not just code changes.

Pass 3 — Connection Strings and File-Based DB Paths

Pattern: sqlite: | :memory: | \.db['")] | new Database( | createConnection(
Files:   **/*.ts, **/*.js, **/*.env, **/*.env.*

SQLite Pattern	Oracle Equivalent
`sqlite:./path/to/db.sqlite`	OCI connection string / TNS alias
`:memory:`	Oracle SGA (in-memory buffering is automatic)
`new Database('file.db')`	`oracledb.getConnection(config)`
`createConnection({ type: 'sqlite', database: '...' })`	`createConnection({ type: 'oracle', ... })`

Pass 4 — Schema Syntax

Pattern: AUTOINCREMENT | INTEGER PRIMARY KEY | TEXT DEFAULT | BLOB | REAL | NUMERIC
Files:   **/*.sql, **/*.ts, **/*.js

SQLite Type	Oracle Equivalent
`INTEGER PRIMARY KEY AUTOINCREMENT`	`NUMBER GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY`
`TEXT` (< 4000 chars)	`VARCHAR2(4000)`
`TEXT` (unbounded / large)	`CLOB`
`BLOB`	`BLOB`
`REAL`	`BINARY_FLOAT` or `NUMBER`
`NUMERIC`	`NUMBER`
`BOOLEAN` (stored 0/1)	`NUMBER(1)` with `CHECK (col IN (0,1))`
`DATETIME` (stored as text)	`TIMESTAMP WITH TIME ZONE`
`AUTOINCREMENT` sequence in DDL	`CREATE SEQUENCE` + trigger, or `GENERATED AS IDENTITY`

Pass 5 — Fallback Logic (Most Commonly Missed)

Pattern: sqlite | SQLite | in.memory | fallback.*db | db.*fallback (case-insensitive)
Files:   **/*.ts, **/*.js

Conditional branches that fall back to SQLite when Oracle is unavailable must be removed entirely — not just replaced.

// Examples to flag:
if (process.env.DATABASE_URL?.includes('sqlite')) { ... }    // [SQLITE_REPLACE]
const db = hasOracle ? oracleDB : sqliteDB;                  // [SQLITE_REPLACE]
const adapter = isProd ? 'oracle' : 'sqlite';                // [SQLITE_REPLACE]

Migration Manifest Format

Output this after all five passes. This is the deliverable for implementation agents.

## SQLite Migration Manifest

| File                       | Line | Pattern Found                           | Marker           | Oracle Equivalent                                        |
|----------------------------|------|-----------------------------------------|------------------|----------------------------------------------------------|
| apps/api/src/db/client.ts  | 3    | `import Database from 'better-sqlite3'` | [SQLITE_REPLACE] | `import oracledb from 'oracledb'`                        |
| apps/api/src/db/schema.ts  | 12   | `id INTEGER PRIMARY KEY AUTOINCREMENT`  | [SQLITE_REPLACE] | `id NUMBER GENERATED BY DEFAULT AS IDENTITY PRIMARY KEY` |
| apps/api/data/state.sqlite | —    | SQLite data file                        | [SQLITE_REPLACE] | Migrate data to Oracle table via INSERT SELECT           |
| apps/api/src/config.ts     | 45   | `sqlite:./data/portal.db`               | [SQLITE_REPLACE] | OCI connection string                                    |
| apps/api/src/app.ts        | 88   | `const db = isProd ? oracle : sqlite`   | [SQLITE_REPLACE] | Remove fallback; Oracle only                             |

**Total touch points: N**
**Estimated implementation tasks: N**
**Data files requiring migration: N**

Each [SQLITE_REPLACE] marker = one atomic implementation unit for Agent 2+.

Common Mistakes

Mistake	Fix
Reading entire files to search	Use Grep with `-n`. Never `cat` a file to scan it.
Skipping Pass 5 (fallback logic)	Fallback branches are the trickiest to remove — always run it.
Assuming `TEXT` → `VARCHAR2` universally	Check actual usage. Content > 4000 chars needs `CLOB`.
Ignoring `.db` data files	Flag them — they require data migration planning, not just code changes.
Missing `drizzle-orm/sqlite-core` schema files	Drizzle's Oracle support differs; all schema files need flagging.
Flagging Oracle-side code as SQLite	Filter false positives — `sqlite` may appear in migration docs or comments.

Arguments

$ARGUMENTS: Optional user-provided target, path, environment, symptom, or constraint. When empty, infer the narrowest safe scope from the current repository context and ask only if multiple high-impact choices remain.

/sqlite-to-oracle-planner

SKILL.md