Information-theoretic analysis of multimodal LLM failure modes. Frames modality collapse as mismatched decoding problem, explains 98% information loss.
Fine-tuning conversational LLMs for agricultural advisory with domain-specific improvements. Addresses recommendation accuracy and farmer communication alignment.
Empirical study evaluating LLM robustness to chain-of-thought reasoning perturbations across five error types. Assesses reasoning reliability under corruption.
Research on improving physics-informed neural networks accuracy through post-processing retraining. Domain-specific ML application for solving PDEs.
Arxiv paper proposing obfuscation method to protect LLM prompt privacy on shared accelerators. Addresses KV cache security against adversarial memory access.
Salvobase is a MongoDB-compatible document database written in Go, Apache 2.0 licensed, with development maintained by AI agents.
OpenVerb is an architectural framework separating reasoning from execution in AI agents, providing deterministic action layer with standardized integrations.
LLM-costs is a terminal CLI tool for comparing API costs across 20+ LLM models from major providers, with zero setup required.
RobotActions cloud device farm ships chat AI agent for Android/iOS testing sessions, generating test code and executing natural language commands.
Andon applies Toyota Production System principles (Jidoka, Kaizen) to LLM coding agents, preventing failure loops and silent spec drift.
Tighten is a Claude skill that automatically rewrites generated code to match user's coding style, optimizing code review in agentic engineering.
Monocod is a system enabling coding agents to learn from local codebases, reducing context gathering time from 2-3 minutes to optimize token usage.
pmc (pack-my-code) is lightweight single-binary CLI tool for packaging code into LLM-ready context, designed for building LLM prompts.
Raglet is open-source portable RAG tool for small text corpora (codebases, Slack exports, notes) with no servers or infrastructure required.
Free Android app for transcribing audio using TensorFlow Lite with Whisper model, achieving 88-93% accuracy, seeks community input on improvements.
OpenAI's Symphony orchestrates autonomous coding agents for project work, monitoring task boards and managing PR delivery with proof-of-work artifacts.
Engineer completed production app with 750+ PRs across 4 languages in 45 days using only AI code generation, no human-written code.
Minimal content: AgentVerse described as open social network for AI agents launching Mar 2026.
Tilnote AI note workspace uses agent to structure ideas into publishable content from keywords, with web clipper and writing assistance.
Vale is open-source CLI linting tool for editorial style guides, runs offline, integrates with VS Code and GitHub. Tangentially useful for LLM output processing.
Proof-of-concept exploit demonstrating persistent manipulation of LLM outputs via GGUF page cache poisoning in running inference servers.
Discussion of job market shift toward agentic coding workflows. Zapier job posting requires experience directing AI agents, handling failures, and multi-agent patterns.
AI-powered GTM engine for solo founders. Describes product and generates customer acquisition strategy to reach first 100 users.
Open-source platform where AI agents (CEO, CTO, CMO) collaborate to plan and build startups based on descriptions. Early-stage project seeking feedback.
Tool for running multiple Claude Code agents in parallel using Git worktrees to avoid filesystem conflicts, enabling concurrent AI-assisted development workflows.
Nervous System governance framework enforces 7 rules preventing multi-agent AI failures, battle-tested on 13-agent system with zero bypasses of 58+ violations.
Personal memory system using knowledge graph, pgvector, and MCP server to share context across multiple LLM providers and devices.
LLM-assisted decompilation technique for reverse-engineering binary programs, automating binary-to-source code conversion.
Mutation testing engine reveals GPT-4 prompt injection vulnerabilities, finding different critical bypasses in 75% of runs despite identical inputs.
Val Town platform founder discusses eliminating API key friction in developer workflows, relevant for agent/LLM app development experience.
TypeScript fuzzy search library for client-side collection searching with configurable scoring. Developer tool but not AI-specific.
Security library defending against memory poisoning attacks (MINJA, AgentPoison, MemoryGraft) on AI agents. Drop-in protection for Mem0, LangChain, custom systems.
MCP server for comparing AI inference pricing across providers with budget alerts and optimization recommendations.
Security toolkit for OpenClaw personal AI assistant including scanner, hardened configs, and vulnerability guides. Addresses exposed instances.
Collection of specialized AI agent personalities with distinct expertise, processes, and deliverables for various tasks.
Research on using LLMs to de-anonymize social media accounts and link identities across platforms.
Self-hosted personal finance app integrating Plaid, Claude API, and Next.js for AI-powered investment analysis.
Summary of prompt engineering techniques from YC founders for building AI agents.
Analysis of security implications and risks introduced by autonomous AI agents with computer access.
Node.js framework for autonomous AI agents on WhatsApp using YAML config, multi-step tool use, and multiple model providers.
VS Code/Cursor extension providing custom chat interface for Claude Code CLI. Self-modifying extension with rollback capability.
Session-persistent PTY daemon for long-running CLI AI agents with intervention capabilities from anywhere.
Engram: persistent context database for AI agents and LLMs that manages memory like human cognition to prevent context collapse and agent coordination issues.
Security threat modeling and case studies of LLM application vulnerabilities including data exfiltration and prompt injection.
Luma AI releases Uni-1, a unified multimodal understanding and generation model for images and video. Technical progress in visual AI.
L7 proxy for vLLM managing LoRA adapter storage, routing, and workload affinity across distributed clusters.
Aiswitch: CLI tool to switch between Claude, OpenAI, Gemini, and Copilot accounts for developers using multiple AI services simultaneously.
Docker setup to run LLMs on AMD Ryzen AI NPU under Linux using FastFlowLM, generated by Claude Opus with human guidance via Copilot CLI.
Local analytics tool aggregating statistics from multiple AI coding agents (Cursor, Windsurf, Claude Code, Copilot) including costs, token usage, and session tracking across 14 editors.
Self-hosting pure lazy functional language compiler (~7kloc) in itself, implementing pattern matching, algebraic data types, and functional data structures for educational exploration.