POP adapters are runtime-facing bindings for canonical persona objects. Their purpose is to project a validated persona object into a runtime-oriented representation without redefining the underlying POP object.
- LangChain adapter scaffold
- CrewAI adapter scaffold
- both are early preview and intentionally dependency-light
- LangChain is the primary execution-layer focus in the current release line
- CrewAI remains secondary and compatibility-focused
- v0.1.5 continues to treat CrewAI as secondary while stabilizing the LangChain execution contract surface
- v0.1.6 continues that line by reducing first-adoption friction rather than expanding helper count
The canonical persona object remains primary. Adapters are projection and binding layers built around that object, and they should preserve its semantics rather than reshape the core schema around runtime preferences.
- not a full runtime integration
- not official LangChain or CrewAI support
- not a guarantee of behavior equivalence across runtimes
Real runtime package integrations may be added later. For the current stage, adapter contract tests are the main proof surface: they show that runtime-facing bindings can exist without changing the canonical POP persona boundary.
CrewAI runtime compatibility should currently be validated on Python 3.10-3.13. Runtime-facing inputs such as tools remain outside the POP persona core even when adapter or execution helpers are used.