Swisper Studio¶

Observability and configuration management for Swisper AI deployments. Swisper Studio provides complete visibility into every LLM call, state change, and tool execution — plus no-code configuration management across environments.

Code repository: Fintama/swisper_studio

Core Capabilities¶

Capability	Description
Complete observability	Every LLM call, state change, and tool execution traced
Visual debugging	Interactive graphs of agent execution
No-code config management	Change LLM models without deployments
Multi-environment	Separate dev, staging, production
Cost tracking	Token usage and spending per trace
Prompt management	Version-controlled prompt editing and deployment
SDK integration	Python SDK for instrumenting Swisper deployments

Architecture¶

See the Architecture Overview for the full system design.

Guides¶

Getting Started¶

Quickstart Configuration — Get up and running quickly
SDK Integration — Full SDK integration walkthrough
Multi-Environment Tracing — Configure tracing across environments

Development¶

Agent Creation — How to create new agents
Prompt Writing — Prompt authoring best practices
Testing — Testing patterns for async database and SQLAlchemy
UI Node Prompt Editing — Editing prompts in the UI
OpenAPI PR Fix — Fixing OpenAPI-related PR issues

SDK¶

SDK Publishing — How to publish SDK releases
SDK Upgrade v0.6.1 — Upgrade guide for v0.6.1
SDK Upgrade v0.5.1 — Upgrade guide for v0.5.1

Operations¶

Setup & Deployment — Complete deployment guide
Configuration Reference — Visual diagrams and connection flows
Swisper Backend Config — Backend configuration details

Decisions¶

Architecture Decisions (ADR) — 9 ADRs covering frontend, backend, SDK, and strategy
TDR-001 Trace Size Optimization — Technical debt tracking

Technology Stack¶

Layer	Technologies
Backend	Python, FastAPI, SQLAlchemy
Frontend	React, TypeScript, MUI
Database	PostgreSQL, Redis (separate instances)
SDK	Python SDK with auto-instrumentation
Infrastructure	Docker, Docker Compose