7 published papers · 6 live products · 10 research initiatives
2.4 million AI agents in production. Zero standard methodology for verifying they work correctly. Agents leak data, exceed budgets, drift from instructions, and fail in ways no one predicted.
Every framework helps you build agents. We make them reliable — with math, not marketing.
| Product | What It Does | Install | Links |
|---|---|---|---|
| agentAssert | Design-by-Contract for AI agents. 12 domain contracts, 293 benchmarks, Θ=0.9541 | pip install agentassert-abc[yaml,math] |
  |
| SuperLocalMemory | Privacy-first AI agent memory. 74.8% LoCoMo, zero cloud, Fisher-Rao retrieval | npm i -g superlocalmemory |
  |
| SkillFortify | Supply chain security for AI agent skills. 22 frameworks, 96.95% F1, 0% FP | pip install skillfortify |
  |
| AgentAssay | Token-efficient regression testing for non-deterministic agents. ~83% cost reduction | pip install agentassay |
 |
| SLM Mesh | P2P communication for AI coding agents. 8 MCP tools, <100ms delivery | npm i -g slm-mesh |
 |
| Qualixar OS | Universal agent operating system. 25 commands, every transport, every IDE | Coming soon |  |
| # | Paper | Venue | Product |
|---|---|---|---|
| 1 | The Living Brain — Biologically-Inspired Forgetting, Cognitive Quantization, Multi-Channel Retrieval | arXiv:2604.04514 | SuperLocalMemory V3.3 |
| 2 | Information-Geometric Foundations for Zero-LLM Enterprise Agent Memory | arXiv:2603.14588 | SuperLocalMemory V3 |
| 3 | Privacy-Preserving Multi-Agent Memory with Bayesian Trust Defense | arXiv:2603.02240 | SuperLocalMemory V2 |
| 4 | Agent Behavioral Contracts — Formal Specification and Runtime Enforcement | arXiv:2602.22302 | agentAssert |
| 5 | Formal Verification for Agent Skill Supply Chain Security | arXiv:2603.00195 | SkillFortify |
| 6 | Token-Efficient Regression Testing for Non-Deterministic AI Agents | arXiv:2603.02601 | AgentAssay |
| 7 | A Universal Agent Operating System | Zenodo DOI:10.5281/zenodo.19454219 | Qualixar OS |
Conference targets: NeurIPS 2026, ASE 2026, AAMAS 2027, ICSE 2027.
| Codename | Focus Area |
|---|---|
| Project Echo | Multi-agent communication degradation benchmarks |
| Project Sentinel | Reliability analysis for AI-generated code |
| Project Rewind | Time-travel debugging for autonomous agents |
| Project Aurora | Chaos engineering for AI agent systems |
# Agent behavioral contracts — 12 domains, 293 benchmarks
pip install agentassert-abc[yaml,math]
# AI agent memory — works with Claude Code, Cursor, 17+ tools
npm install -g superlocalmemory
# Agent skill security — scan 22 frameworks in one command
pip install skillfortify && skillfortify scan
# P2P agent communication — agents discover + message each other
npm install -g slm-mesh && slm-mesh start- Every tool is backed by published, peer-reviewed research — not blog posts
- We solve problems with mathematical proofs — not heuristics
- Every product is open-source and framework-agnostic
- Privacy-first: your data stays on your machine
| Site | What's There |
|---|---|
| varunpratap.com | Author portfolio — all products, 7 papers, blog |
| agentassert.com | agentAssert — benchmarks, contracts, getting started |
| superlocalmemory.com | SuperLocalMemory — architecture, integrations, research landscape |
| qualixar.com | Platform hub — all products, documentation |
- Read our papers on arXiv
- Star the repos you find useful
- Open issues for bugs or feature requests
- Cite our work in your research (BibTeX on each repo)
An independent research initiative by Varun Pratap Bhardwaj
Senior Manager & Solution Architect at Accenture · 15 years across 6 industries
We don't just identify problems in agent development. We prove solutions.