What to use for new integrations
- Create and store API keys with the
sk-planck-orpk-planck-prefix. - Use
PLANCK_API_KEYfor client and SDK configuration. - Use
Planck-*request headers. For example, usePlanck-Auth,Planck-User-Id, andPlanck-Property-*. - Route explicitly through the
planckmodel-provider slug. - Use
/v1/planck-dataset,/v1/planck-sql, and/v1/playground/requests-through-planckfor the renamed API resources. - Read
Planck-*response and webhook headers, such asPlanck-IdandPlanck-Signature.
Compatibility behavior
- Existing
sk-bifrost-andpk-bifrost-credentials remain valid. Bifrost-*request headers remain accepted. If both names are present, the correspondingPlanck-*header wins.- Compatibility response and webhook headers are emitted alongside the Planck names during the migration period.
- The
bifrostmodel-provider slug resolves to the canonicalplanckprovider. - The corresponding
/v1/bifrost-dataset,/v1/bifrost-sql, and/v1/playground/requests-through-bifrostroutes remain accepted. - New JavaScript integrations use the
@inquantum/planck-*package family, including@inquantum/planck-helpers,@inquantum/planck-async, and@inquantum/planck-mcp. Python integrations use the correspondingplanck-helpersandplanck-asyncdistributions. - Internal monorepo workspaces and compatibility identifiers may retain their
@bifrost/*names when changing them would break internal consumers. - Existing Docker automation can continue to invoke
bifrost-compose.shand thebifrostprofile. New automation should invokeplanck-compose.shand theplanckprofile.