React hook for using self-correcting setInterval, augmented by management methods (start, stop, isActive)

Self-correcting memory for LLM agents. A Claude Code plugin that learns from sessions, surfaces relevant memories, measures whether they're actually followed, and fixes the ones that aren't.

A deterministic pipeline that diagnoses, explains, and safely auto-fixes failures in LLM agent systems.

GraphNews is a multi-agent system that acts as a self-correcting editorial team, generating newsletters

Adaptive vector search with self-correcting embeddings. Fixes semantic collapse in RAG systems via spectral chelation, dynamic dimension masking, and neural adaptation.

🧬 ScienceProves.Me - A RAG-powered assistant that answers questions with high precision using scientific evidence.

Self-Auditing AI: a self-correcting algorithm that holds a checkable invariant about its own output, re-tests and escalates on deviation, and records a full audit trail. Real scipy root-finder + sensor anomaly detection.

CodegniPy is a groundbreaking Python library that elevates AI to a first-class citizen of the language. It introduces a cognitive computing engine where deterministic code and non‑deterministic LLM reasoning coexist seamlessly, enabling you to write Python with intent rather than just instructions.

Agentic forecasting application with autonomous model selection, multi-indicator prediction, and self-correcting pipelines

Self-correcting Python coding agent: plan → generate → execute tests → fix → repeat. LangGraph loop with sandboxed subprocess execution. Supports Claude and GPT-4o.

The simplest autonomous AI-agent loop that provably halts: goal, an unfakeable check, three hard stops, durable memory, in ~110 lines of bash. A foundation for loop engineering with Claude or any coding agent.

Comparative study of two Agentic AI architectures for automated data science: hidden-tool agents vs transparent code-generating agents. Built with CrewAI, OpenAI GPT-4o, tested on Titanic & House Prices datasets.

📰 Automate news research and curation with GraphNews, a multi-agent system that ensures high-quality, factual content through self-correction.

An autonomous AI agent that writes Python code, catches its own errors, and self-corrects.

A self-correcting multi-agent system that audits and actively fixes Kaggle notebooks. Powered by LangGraph and Gemini 2.5, it uses an autonomous feedback loop to improve code quality, documentation, and reproducibility scores in minutes.

Runtime harness for AI agents — a TypeScript framework that plans, executes, and verifies its own work via a Planner → Executor → Evaluator loop.

Self-correcting 5-pillar financial AI with continuous improvement cycle. Production: boseongkang.github.io/newstrend