Releases: inference-stack-llc/electripy-studio
Releases · inference-stack-llc/electripy-studio
v0.1.3
v0.1.2
Fixed
- Repository map and development workflow images now use absolute URLs for PyPI rendering.
v0.1.1
Fixed
- README social-preview image now uses absolute URL so it renders correctly on PyPI.
v0.1.0 — Initial Release
ElectriPy Studio v0.1.0
Production-grade Python toolkit for AI product engineering.
Highlights
- Core: Configuration, structured logging, typed error hierarchy
- Concurrency: Retry (sync/async) with exponential backoff, async token-bucket rate limiter, task groups
- I/O: JSONL read/write utilities
- CLI: Typer-based interface with
doctor,rag eval, anddemo policy-collabcommands - AI — LLM Gateway: Provider-agnostic sync/async clients with structured output helpers and request/response policy hooks
- AI — Policy Gateway: Regex-based detection, sanitization, deny/allow/require-approval across preflight/postflight/stream/tool stages
- AI — Agent Collaboration Runtime: Bounded multi-agent orchestration with hop limits and deque-based message routing
- AI — RAG Eval Runner: Dataset loader, evaluation runner, CLI benchmarking
- AI — Telemetry: Provider-agnostic observability (JSONL, optional OpenTelemetry)
- AI — Product Engineering: Streaming chat, agent runtime, RAG quality/drift, hallucination guard, response robustness, prompt engine, token budget, context assembly, model router, conversation memory, tool registry
Code Quality
frozen=True, slots=Truedataclasses for all domain value objects__slots__on all service classes__all__exports in every moduleSequence/Mappingin all Protocol signatures- 168 tests passing | ruff | black | mypy | mkdocs --strict
Community
- CONTRIBUTING.md, CODE_OF_CONDUCT.md, SECURITY.md, CHANGELOG.md
- GitHub issue templates (bug report, feature request)
See the full CHANGELOG for details.