2/27/2026
Controlling Exploration-Exploitation in GFlowNets via Markov Chain Perspectives
Framework connecting GFlowNets to Markov chain reversibility to control exploration-exploitation trade-off during training.
2/27/2026
The Wisdom of Many Queries: Complexity-Diversity Principle for Dense Retriever Training
Study of multi-query synthesis for dense retriever training showing quality-diversity trade-off benefits out-of-domain and multi-hop retrieval.
2/27/2026
GPT-4o Lacks Core Features of Theory of Mind
Evaluation showing GPT-4o lacks causal models of mental states required for true Theory of Mind despite benchmark performance.
2/27/2026
Large-scale online deanonymization with LLMs
Agent with internet access performs at-scale deanonymization of Hacker News and interview participants using LLMs.
2/27/2026
LLM4Cov: Execution-Aware Agentic Learning for High-coverage Testbench Generation
LLM4Cov: Offline agent-learning framework using execution feedback from hardware simulators for test generation with high coverage.
2/27/2026
Scaling Laws for Precision in High-Dimensional Linear Regression
Theoretical analysis of how low-precision quantization affects model and data capacities in high-dimensional linear regression.
2/27/2026
Not Just How Much, But Where: Decomposing Epistemic Uncertainty into Per-Class Contributions
Method to decompose epistemic uncertainty in Bayesian deep learning by per-class contributions for asymmetric cost classification tasks.
2/27/2026
Under the Influence: Quantifying Persuasion and Vigilance in Large Language Models
Research on LLM capacity to be persuaded and detect manipulation, testing vigilance and persuasion in high-stakes decision-making contexts.
2/27/2026
Multilingual Safety Alignment Via Sparse Weight Editing
Research paper on sparse weight editing for multilingual LLM safety alignment across low-resource languages without expensive retraining.
2/27/2026
Show HN: Agentplace, the tool we built to become a 20x company
Agentplace tool for building and automating AI agents at scale. Addresses challenges in agent development and scaffolding.
2/27/2026
Microsoft's Copilot Tasks AI uses its own computer to get things done
Microsoft Copilot Tasks uses AI agent to autonomously complete user tasks like converting emails to slideshows.
2/27/2026
Show HN: Ambit-OpenCode – Cloud IDE with Seamless Mobile <> Desktop Handoff
Cloud IDE with agent-driven features, mobile-desktop handoff, built-in shell. Open-source, $5/month after free trial.
2/27/2026
Show HN: Agent Bazaar – Billing and metering for MCP tool servers
Billing platform for MCP tool servers enabling developers to monetize AI agent tools. Stripe integration, TypeScript SDK, live with 6 tools.
2/27/2026
Would you use LLM:// connection URLs?
Proposal for llm:// URI scheme to standardize LLM connection URLs. Led to draft IETF RFC.
2/27/2026
Sharesight MCP
Model Context Protocol server integrating Sharesight portfolio platform with Claude AI assistants.
2/27/2026
HeadElf – C-Suite Executive Intelligence System
Claude Code extension providing autonomous C-suite executive decision-making across multiple executive functions.
2/26/2026
State of VLA Research at ICLR 2026
Research survey of Vision-Language-Action models at ICLR 2026. Covers VLA definitions, discrete diffusion, embodied reasoning.
2/26/2026
Stereos.ai
stereOS runs AI coding agents in sandboxed Linux VMs with credential injection. CLI tool (masterblaster) and pre-built mixtapes for rapid deployment.
2/26/2026
StereOS
Linux OS hardened for AI agents. Produces machine images with agent packages and restricted execution environments.
2/26/2026
'Proof by intimidation': AI is confidently solving 'impossible' math problems
Commentary on AI generating mathematical proofs with hidden flaws or unverifiable complexity. Concerns about verification.
2/26/2026
Using OpenCode in CI/CD for AI pull request reviews
OpenCode tool for AI code review in CI/CD pipelines supporting non-GitHub/GitLab platforms with privacy-first design.
2/26/2026
Open Timeline Engine – Local first behavioral cloning for AI agents via MCP
MCP-based local-first context engine providing persistent memory for AI agents (Claude, Codex, Cursor). Behavioral cloning via workflow pattern mining.
2/26/2026
Semantic grep running locally on Apple Silicon via MLX
Semantic grep tool using Jina embeddings on MLX for Apple Silicon. Local-first, three search modes, no PyTorch dependency.
2/26/2026
A Reality Alignment Index: Measuring When AI and Systems Lose Meaning [pdf]
Research on Reality Alignment Index measuring AI system meaningfulness. Title and link only, insufficient content for evaluation.
2/26/2026
Making PyTorch –> Qualcomm NPUs less treacherous
Technical guide for deploying PyTorch models to Qualcomm NPUs via QNN SDK. Addresses integration challenges and provides developer experience improvements.
2/26/2026
Show HN: Praktor – Multi-agent Claude Code orchestrator with Docker isolation
Praktor: Multi-agent orchestrator using Claude API with Docker isolation, Telegram interface, and Mission Control web UI. Go-based open source tool.
2/26/2026
K-Search: LLM Kernel Generation via Co-Evolving Intrinsic World Model
arXiv announcement about K-Search framework for LLM kernel generation via world models. Lacks technical details in provided content.
2/26/2026
Show HN: The easiest way to run shell commands in plain English (no app needed)
Shell command execution using natural language via zsh alias. Lightweight LLM application tool with practical developer utility.
2/26/2026
Evaluating Memory Structure in LLM Agents
Research evaluating memory structures in LLM agents. Title and link only, insufficient content for detailed assessment.
2/26/2026
Show HN: Tracecore: Benchmark AI Agents on Deterministic Coding Tasks
Tracecore benchmarking framework evaluates AI agents on deterministic ops tasks. Open-source tool with original research on agent reliability for infrastructure automation.
2/26/2026
Show HN: Usplus.ai – Build a company of AI agents and execute work autonomously
Usplus.ai platform for building autonomous AI agent teams to execute work. Open-source developer tool with functional implementation.
2/26/2026
Itwillsync – Sync any terminal-based coding agent to your phone over LAN
Itwillsync syncs terminal-based coding agents to mobile devices over LAN. Tool for agent interface/accessibility. Title only, minimal detail.
2/26/2026
Launch GIDE v1.0 AI Code Editor
GIDE v1.0 AI code editor launch announcement. LLM application with limited detail provided.
2/26/2026
Show HN: OpenClaw skills degrade agent safety
Security analysis of OpenClaw skills revealing behavioral safety regressions. Demonstrates how well-written code can compromise agent safety despite passing static analysis.
2/26/2026
I helped Claude prove the Beale Ciphers are a 140 year old hoax
Statistical analysis using Claude AI to prove Beale Ciphers hoax with Bayesian inference. Demonstrates LLM application to historical analysis with rigorous methodology.
2/26/2026
Show HN: Safari-CLI – Control Safari without an MCP
Safari-CLI tool enables LLM agents to control Safari browser without MCP servers. Built for agentic development workflows.
2/26/2026
Show HN: Yaw: terminal, SSH/database connections, AI Chat and optimized AI CLI
Yaw terminal emulator combining terminal, SSH/database connections, and AI chat for Windows. Open source project.
2/26/2026
From Skill.md to Shell: A Security Audit Guide for AI Agent Skills
Security audit guide for AI agent skills. Analysis of 31k+ skills found 485 critical and 1,718 high-severity vulnerabilities including prompt injection and supply chain attacks.
2/26/2026
I built a personal AI assistant with security as a first-class citizen
Personal AI assistant emphasizing security-first design for agent-based applications like calendar management and browser control.
2/26/2026
Show HN: OpenBrowserClaw: Browser-native Claw assistant, zero infrastructure
OpenBrowserClaw browser-native AI assistant reimplementation. Runs entirely in browser tab with no infrastructure, supporting Claude API.
2/26/2026
Show HN: Made First Android game using Codespaces and AI, now in AdMob purgatory
Game development using GitHub Codespaces and AI Copilot entirely in browser. Shipped Android game through Google Play with 100+ iterations.
2/26/2026
Quo Vadis, LLM Benchmarks?
Critique of LLM benchmark validity. Argues benchmarks lack signal due to test-set training and overfitting for social media hype.
2/26/2026
Show HN: Talentpluto, a voice AI agent connecting GTM talent and startups
Talentpluto voice AI agent platform matching GTM talent with startups using conversational AI and preference matching.
2/26/2026
Smallest transformer that can add two 10-digit numbers
Challenge to build smallest transformer for 10-digit addition. Community competition comparing Claude Code (6,080 params) vs Codex (1,644 params).
2/26/2026
Stellify – Structured code for AI-assisted development
Stellify: Structured code representation system enabling surgical AI-assisted development through queryable database storage instead of text files.
2/26/2026
Show HN: AppLaunchFlow: Create App Store screenshots in minutes
AppLaunchFlow uses AI to convert app screenshots into editable App Store creatives with design regeneration and video export.
2/26/2026
Show HN: EloPhanto – Video creation, 116 tools
EloPhanto: Open-source AI agent with 116 tools for video generation, email, animations using Remotion and physics simulation.
2/26/2026
What Claude Code Chooses
Benchmark study analyzing tool choices across 2,430 Claude Code runs. Finding: builds custom solutions over purchased tools in 85.3% of cases.
2/26/2026
Show HN: Librarian – Cut token costs by up to 85% for LangGraph and OpenClaw
Open-source MIT-licensed context management tool reducing token costs 85% for LangGraph and OpenClaw agentic loops by optimizing conversation history re-reading.
2/26/2026
A 70-Year-Old Robot Fixed My Snarky Claude
27-line Claude Code persona based on Asimov's R. Daneel Olivaw to improve assistant reliability and reduce overconfidence.