​

Data classification guidance

Before placing information into Starfire, decide what kind of data it is and whether the selected context is appropriate for it.

​

Practical categories

​

Public

Information intended for unrestricted public distribution. Examples: published documentation, public marketing copy, open-source code.

​

Internal

Business information intended for authorized team use but not necessarily public. Examples: internal procedures, draft plans, non-public project notes.

​

Confidential

Information whose exposure could cause business, customer, contractual, or privacy impact. Examples: unpublished product plans, private source code, customer records, sensitive financial information.

​

Secret / credential

Information that directly grants access. Examples: passwords, API keys, tokens, private keys. Do not place secrets into normal Starfire AI context.

​

Choose the context

​

Durable vs temporary context

A confidential file attached to one chat and the same file indexed into organization Knowledge have different reuse and access implications. Use the narrowest lifetime and audience required by the task.

​

Organization policy

Organizations can apply their own data-handling rules in addition to Starfire product permissions. Team policy should decide which categories are permitted in Projects, Knowledge, integrations, or AI workflows.

​

Developer integrations

When an API or webhook sends data from Starfire into another system, the receiving system’s security and retention rules matter too.