CARE Platform
The CARE Platform is the reference implementation of the CARE specification. What it is NOT: a monitoring layer bolted onto existing agent frameworks. The CARE Platform provides the Trust Plane and Execution Plane as architecturally separate systems. Governance decisions and execution decisions operate in structurally distinct code paths, not as annotations on the same process.
What it does
Section titled “What it does”The CARE Platform implements the full governance architecture for autonomous agent systems:
- Trust Plane management: Define and manage constraint envelopes across the five dimensions: Financial, Operational, Temporal, Data Access, Communication
- Execution Plane orchestration: Agent task execution within human-defined boundaries, with automatic constraint verification at every node
- Trust verification: Integration with EATP for cryptographic trust lineage from organizational authority to individual agent actions
- Observation infrastructure: Visibility into agent execution patterns, boundary approaches, and escalation events
- Constraint evolution: Tools for refining constraint envelopes based on observed execution patterns, the reflective improvement loops that CARE requires
Architecture
Section titled “Architecture”The platform implements the three-part architecture defined in the CARE specification:
- Trust Plane: human-owned configuration of values, boundaries, and accountability chains
- Execution Plane: AI agent operations within defined constraint envelopes
- Trust Verification Bridge: the mechanism connecting execution to trust, implementing the four verification categories (auto-approved, flagged, held, blocked)
Project details
Section titled “Project details”| License | Apache 2.0 |
| Language | Python |
| Source | github.com/terrene-foundation/care |
| Specification | CARE (CC BY 4.0) |
| Owner | Terrene Foundation |
What the CARE Platform demonstrates
Section titled “What the CARE Platform demonstrates”The platform is a working proof of concept showing that the Constrained Organization model can be implemented in software:
- Architectural separation is enforceable: Trust decisions and execution decisions operate in structurally distinct systems
- Constraint envelopes are composable: Five-dimensional boundaries can be defined, combined, and verified at machine speed
- Trust lineage is verifiable: Every agent action traces back to human authority through cryptographic proof
- Knowledge compounds: The platform’s five-layer CO architecture persists institutional knowledge across sessions
Relationship to other projects
Section titled “Relationship to other projects”The CARE Platform builds on and integrates other Foundation projects:
- EATP SDK: Provides the cryptographic trust lineage infrastructure
- Kailash Python SDK: Provides workflow orchestration (140+ nodes, runtime execution, constraint propagation)
- CO specification: Defines the five-layer knowledge architecture the platform implements
Getting started
Section titled “Getting started”See the GitHub repository for installation instructions, documentation, and examples.
For a lighter introduction to the trust protocol without the full platform, see the EATP SDK and the Quickstart guide.