CAP is an open protocol for capability disclosure, invocation, and provenance in causal reasoning systems.
This repository contains the CAP normative specification, supporting schemas, reader documentation, and the source for causalagentprotocol.io.
- capability discovery
- causal verb semantics
- conformance levels
- assumptions disclosure
- provenance and transport bindings
- how a causal engine discovers graphs
- how data is ingested, stored, or refreshed
- agent orchestration
- implementation-specific discovery surfaces
- which causal engine is scientifically correct
- Overview
- Getting Started
- Quickstart for Clients
- Quickstart for Servers
- Normative Specification
- Schemas
- Contributing
- Governance
specification/contains the protocol contractdocumentation/contains onboarding, concepts, guides, and rationaleschema/contains machine-readable artifacts and minimal examplessite/contains the website source and presentation layer
- current spec version:
0.2.2-draft - Level 3 counterfactual remains reserved in
v0.2.x - where draft prose and current adapter behavior differ, the repo records the gap explicitly rather than hiding it
- minimal capability-card examples live in
schema/examples/ - example implementations and adapters are useful references, but they are not the CAP standard itself
- the site should source content from this repo rather than becoming a second content fork