FAQ
General
What is MPLP?
MPLP is a vendor-neutral lifecycle protocol for AI agent systems.
What problem does MPLP address?
MPLP addresses protocol-level concerns such as:
- lifecycle structure
- observability
- governance hooks
- portability across implementations
These are protocol properties, not guarantees of any specific runtime or product.
Is MPLP open source?
Yes. MPLP repository, docs, and related public package surfaces are published under Apache 2.0.
Architecture
What are the protocol layers?
| Layer | Scope |
|---|---|
| L1 | core protocol objects and schema layer |
| L2 | coordination/module semantics |
| L3 | runtime realization layer |
| L4 | integration boundary layer |
L3 and L4 should not be read as mandatory runtime products.
What is PSG?
PSG is a runtime-side concept used in some implementation guidance. It is not a protocol-core object and should not be used as the primary entry point for reading MPLP semantics.
Implementation
Can I modify the schemas?
The frozen protocol line is protocol_version: 1.0.0.
Use the repository versioning and governance records for current change policy.
How do I verify my implementation?
Start with:
- Entry Points
- Specification
- Golden Flows
- Validation Lab Overview when Lab-side adjudication context is needed
Which package surfaces are available?
| Surface | Role |
|---|---|
@mplp/schema | direct schema/data mirror |
@mplp/sdk-ts | TypeScript facade helper |
@mplp/runtime-minimal | runtime-minimal helper |
mplp-sdk | Python protocol helper |
Availability does not imply endorsement or completeness beyond the published surface itself.
Support
Where can I report bugs?
GitHub Issues: https://github.com/Coregentis/MPLP-Protocol/issues
How do I contribute?
Start with the repository governance source tree and the docs-side governance
helper pages under /docs/evaluation/governance.