Back to Home
· product · apcore
apcore
Governed, protocol-neutral capability runtime
apcore defines machine-readable application capabilities and enforces their execution contract. Required schemas, behavioral annotations, ACL, approval gates, middleware, call-chain guards, structured errors, and trace context run behind one registry and executor. Maintained Python, TypeScript, and Rust SDKs follow the same normative protocol while adapters project modules to MCP, A2A, CLI, HTTP, or direct code.
· live runtime
Run apcore-js 0.26 in your browser
Execute real schema validation and approval behavior using the browser-safe SDK entry.
· 01
Features
01Cognitive Interface: Intent-oriented contracts for the AI-native era
02AI Collaboration Lifecycle: Discovery, Strategy, Governance, and Recovery stages
03Directory as ID: Convention-based module discovery from filesystem hierarchy
04Structured Error Guidance: Optional metadata callers may use to choose a recovery path
05Three-Layer Metadata: Core (Schemas), Annotation (Behavior), and Extension (Free Metadata)
06Unified APCore Client: Single entry point for registry, executor, events, and control
07Schema-Driven: Mandatory input/output contracts with LLM-specific extension fields
08Approval System: Secured human-in-the-loop enforcement for sensitive operations
09Observability: ErrorHistory and UsageCollector provide bounded runtime evidence
10Cross-Language SDKs: Python, TypeScript, and Rust follow the same normative protocol
· 02
Get Started
Python · Implementation
Python reference implementation of the apcore specification.
Install
$ pip install apcoreQuick Start
python.pyfrom apcore import APCore
client = APCore()
@client.module(id="math.add", description="Add two numbers")
def add(a: int, b: int) -> dict:
return {"sum": a + b}
result = client.call("math.add", {"a": 10, "b": 5})· related
