agentic-workflows-orchestration

A self-hosted, dockerized agentic coder with Kilocode and Codex support

An example application that exposes a multi-step, quality-controlled LLM workflow built with LangGraph through a FastAPI service.

Kubernetes operator that manages AI agents as first-class resources

Stop AI from lying with tests. A family of Claude Code skills enforcing true test-first TDD via strict agent boundary isolation and spec-implementation splitting.

Agentic flow for scraping data using SearXNG.

Production-grade agentic research system with schema-enforced verifiable citations to eliminate AI hallucination. Built with LangGraph & FastMCP.

A repository-based workflow for coordinating AI development tools using deterministic, snapshot-driven shared context.

LangGraph multi-agent orchestrator template — dynamic Supervisor routing, 5 example specialist agents, and optional human-in-the-loop review. POC tier of the Multi-Agent Orchestrator Template family: in-memory state, no auth, minutes to first run.

LangGraph multi-agent orchestrator with per-user JWT auth, SQLite-backed durable run state, and a queryable audit trail. Intermediate tier of the Multi-Agent Orchestrator Template family.

Microsoft Fabric & Power BI development workspace in VS Code: a 47-agent GitHub Copilot organisation with a strict coordinator, embedded Fabric skills, tool-aware CLI/MCP automation, and a governed DEV-to-PROD workflow.

Meta-planner for Claude Code skills, harnesses, and CLI tooling — semver-versioned plans, an approve gate, orchestrated builds. 4 commands (intake, --init, --suggest, --build), 3 lanes (planned / in progress / done), one commit per phase. Installs as a skill; plans live in your repo. MIT.

Local development CLI for running AI agent pipelines without a Kubernetes cluster

MedGemma Clinical Robustness Assistant Multi-agent dermatology diagnosis system with safety-governed agentic pause and evidence-grounded RAG (Kaggle Med-Gemma Impact Challenge).

Langgraph-powered agent that runs a coding task in gVisor

AI-driven career path optimization and skill-mapping utilizing LLMs, RAG architecture, and Vector Embeddings

An append-only ledger convention (HITL.md) to close the loop with AI coding agents by logging human rulings on taste, scope, and when to ask.

An enterprise-grade Agentic Pipeline for customer support. Replaces unpredictable autonomous agents with a deterministic state-machine, constrained LLM reasoning (Groq/Llama-3), and strict Python guardrails to guarantee zero hallucinations in financial actions.

Helm charts for kubeswarm - AI agents as Kubernetes-native resources

Senior AI Engineer @ Coinbase | Architecting Enterprise LangGraph & RAG Systems | Bridging Philosophy & Distributed Systems

LangGraph implementation of ReWOO: Reasoning Without Observation framework