Skip to content
Conduit

skill

agent-orchestration-framework

An orchestration framework with specialized AI agents collaborating under centralized coordination.

HopeyCodeDS/agent-orchestration-frameworkcommunitylistedunreviewed
Use skill

Set up

Skills are installed by placing their folder where your agent loads skills.

Download ZIP from GitHub Open the repository

The download comes from github.com/HopeyCodeDS/agent-orchestration-framework itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 23, 2026

Agent Orchestration Framework

A production-grade multi-agent AI orchestration framework where specialized agents (validation, retrieval, reasoning, verification) collaborate under centralized coordination to solve complex tasks.

Unlike simple prompt chains, this system implements explicit state machine orchestration, per-agent circuit breakers, full OpenTelemetry observability, and an automated evaluation harness.

Architecture

Client -> API Gateway (FastAPI) -> Orchestrator (State Machine)
                                      |
                    +--------+--------+--------+
                    |        |        |        |
               Validation  Retrieval  Reasoning  Verification
                Agent       Agent      Agent      Agent
                    |        |        |        |
              [Rules DB]  [Vector DB] [Memory]  [Ground Truth]

Each agent has: typed request/response contracts, confidence scoring, circuit breaker protection, and health probes.

Tech Stack

CategoryTechnology
RuntimePython 3.10+
APIFastAPI, uvicorn
ValidationPydantic v2
LLM Gatewaylitellm (provider-agnostic)
Vector DBChromaDB (dev) / Qdrant (prod)
ObservabilityOpenTelemetry + structlog
Configurationpydantic-settings + YAML
Testingpytest, deepeval, respx

Quickstart

# 1. Create and activate virtual environment (Python 3.10+)
python -m venv .venv

# Windows (PowerShell)
.\.venv\Scripts\Activate.ps1

# macOS/Linux
# source .venv/bin/activate

# 2. Install in editable mode with dev tools
pip install -e ".[dev]"

# 3. Run formatters, linters, and tests
black . && ruff check . --fix && mypy . && pytest

Project Structure

agents/           # Agent implementations (extend BaseAgent)
orchestrator/     # State machine, circuit breaker, retry policy
observability/    # OpenTelemetry tracing, structlog, metrics
evaluation/       # Automated evaluation harness + scoring
api/              # FastAPI routes, auth, health checks
config/           # YAML configs + pydantic-settings
tests/            # unit/, integration/, e2e/

Key Design Decisions

  • State machine over prompt chaining — deterministic testing, audit trails, controlled failure recovery
  • Async-first API — LLM calls are never awaited in route handlers; clients poll for results
  • Circuit breaker per agent — isolated failure domains; one unhealthy agent doesn't cascade
  • Hashed payloads in telemetry — PII protection; no sensitive data in log aggregators
  • litellm as LLM gateway — provider-agnostic with automatic fallback chains
↑↓ navigate openesc close