Twin Memory

CodeTwin tracks execution context and learned patterns locally so future runs can stay consistent with prior decisions and constraints.

Memory is stored in the local CodeTwin SQLite data store on your machine.

What gets stored

CodeTwin memory graph includes node types such as:

• decision
• constraint
• failure
• task
• file
• context

This supports better continuity across sessions.

How it is used

• Reuses prior implementation context for similar tasks.
• Preserves user/team preferences across runs.
• Avoids repeating known failure paths.
• Improves consistency of edits in the same project.

Inspecting related data

While there is no dedicated memory show/reset top-level command in this build, you can inspect related session history and usage with:

codetwin session list
codetwin export [sessionID]
codetwin stats

Privacy

Memory and credentials are stored locally by default. Keep repository-level config in source control if desired, and keep secrets in local credential storage or environment variables.

Operational tip

Use separate projects or sessions when you want strict context separation between different codebases.