sdk-python

ai-sdlc

Python SDK for the AI-SDLC Framework — a Kubernetes-style declarative framework for governing AI agents in software development lifecycles.

Installation

pip install ai-sdlc

Quick Start

from ai_sdlc.core.types import Pipeline, API_VERSION
from ai_sdlc.core.validation import validate_resource
from ai_sdlc.builders.builders import PipelineBuilder

# Build a pipeline using the fluent API
pipeline = (
    PipelineBuilder("my-pipeline")
    .add_trigger({"event": "issue.assigned"})
    .add_provider("github", {"type": "github"})
    .add_stage({"name": "implement"})
    .build()
)

# Validate against JSON Schema
result = validate_resource(pipeline.model_dump(by_alias=True))
assert result.valid

Modules

Module	Description
core	Pydantic models for all 5 resource types, JSON Schema validation, comparison, provenance
builders	Fluent builder classes for resource construction
policy	Enforcement engine, autonomy evaluation, complexity routing, authorization
adapters	Interface Protocols, adapter registry, community stubs
reconciler	asyncio-based reconciliation loop with domain reconcilers
agents	Orchestration patterns, executor, multi-tier memory
security	Sandbox, JIT credentials, kill switch, approval workflow Protocols
telemetry	OpenTelemetry semantic conventions, structured logging
compliance	Regulatory framework mappings (EU AI Act, NIST AI RMF, ISO 42001, etc.)
metrics	Metric store, standard metric definitions
audit	JSONL audit logging with tamper-evident hashing

Requirements

• Python 3.11+
• pydantic >= 2.0
• jsonschema >= 4.20
• PyYAML >= 6.0
• opentelemetry-api >= 1.20

License

Apache-2.0