LEASH — Lightweight Encrypted Agent Secret Handling. A proposed companion standard to MCP for secure, auditable, human-controlled secret access by AI agents.
Code for the ICML 2026 position paper arguing that text embeddings should capture implicit semantics, not just surface meaning.
Code for constructing and evaluating neuro-symbolic prompting methods. Used for experiments in "The Road to Generalizable Neuro-Symbolic Learning Should be Paved with Foundation Models."
Vocabulary anchor for the Personal AI OS thesis — kernel-layer substrate for human-AI co-evolution.
This repository conducts security audits on digital identity verification systems, identifies vulnerabilities through whitepapers and technical POCs, and offers decentralized, privacy-preserving solutions.
A position paper on the missing kernel layer between human judgment and self-evolving AI agents — provenance, drift protection, co-evolution.
Governed multi-model concurrence orchestration - debate until concurrence, not forced consensus. Position paper / thesis.
The Limits of My Tokens: The Token-Substrate Hypothesis and the Coinage Probe — a position paper on LLM cognition paired with a pre-specified multi-model probe study.
Position paper: Elastic Automators — A Diagnostic Vocabulary for Language-Model-Driven Workflow Systems. Author: Alexandru Mares (allemaar.com).
Agents instrumented with drift detection — execution-overrun, abstraction-drift, premature closure surfaced.
Position paper: coding agents need an explicit model of codebase decisions, not just structure"
[NeurIPS 2025, Position Paper Track] Simulating Society Requires Simulating Thought. LLM social simulation is stuck at demographics in, behavior out. We argue for reasoning fidelity: auditable causality, grounding in real individuals, consistency under counterfactuals. Project page and interactive demo.
Provenance-enforced calibration: external feedback → tagged source → schema delta → human ratifies → audit trail.
Self-evolving agents whose every mutation must emit provenance — speed preserved, silent drift eliminated.
Kernel implementation of Personal AI OS — runtime substrate enforcing provenance and judgment integrity.
Position paper and white paper drafts for open, modular, local-first AI for Science workflows.
Agents that distinguish current judgment from stale opinion in persistent memory.
Position paper: The Pre-Articulation Observability Boundary - A Structural Constraint on Language-Based AI Systems (December 2025)
Companion code and data for a neurosymbolic AI position paper on taxonomy, probing, and symbolic reasoning benchmarks.
Position paper arguing SWE-Bench's task design optimizes coding agents for isolated bug fixing, not real software engineering
To associate your repository with the position-paper topic, visit your repo's landing page and select "manage topics."