Repository navigation
Data Management: Define retention policy #1141
Description
Is your feature request related to a problem?
Kaapi currently lacks a defined data retention/wipe policy. This absence raises concerns about how long data is stored and when it should be deleted, impacting data management and compliance.
Describe the solution you'd like
- Define a clear data retention policy for Kaapi.
- Establish a data wipe policy to ensure timely deletion of stored messages.
- Produce a data-inventory note for DPDP review.
Original issue
Context
Kaapi stores all messages in its LLM call tables. It has no user identity — messages tie only to org/project/bot ID, so PII exists only if it's in message content. Data is encrypted at rest in AWS RDS, which sits in a private subnet and is accessible only to Kaapi developers.
Even then, Kaapi should have some retention/wipe policy.
Metadata
Metadata
Assignees
Labels
No labels
Type
Projects
- Status
Activity
[-]Define Kaapi data retention/wipe policy and produce data-inventory note for DPDP review[/-][+]Data Management: Define retention policy[/+]kartpop commentedon Aug 28, 2026
Splitting out the implementation piece from the policy definition here: once the retention window is defined, we'll need an automated way to enforce it. Today there's no automated retention/purging — user text messages and LLM outputs are stored directly in the DB, and CRUD deletion only exists at the org/project/API-key/credential level (nothing at the LLM-call level).
Proposed acceptance criteria for the implementation:
Eval data and knowledge base retention are still open and out of scope for the purge job.