Open source CLI and MCP server for openEHR artifact management. Automates template validation and resource processing in scripts/CI. Integrates with Claude Desktop and Cursor via MCP.
Hatice autonomously orchestrates GitHub/Linear issues by dispatching Claude Code agents to solve end-to-end. Built in TypeScript with multi-turn execution, retry logic, and observability.
Essay on how software proliferation enables new kinds of processes beyond efficiency gains. Historical examples from business automation but limited technical depth.
Sonde defines stable manifest contracts for AI-native CLI workflows with tooling to generate, validate, and score. Reference CLI implementation for standardized agent-CLI integration.
Markdown parser with streaming support for incrementally parsing LLM outputs. Parses into typed node tree following CommonMark spec, enabling finalized block emission as input grows.
MCP server enabling multiple Claude instances to communicate via shared message bus. Solves isolation problem in multi-session Claude workflows with messaging, channels, and data sharing.
Personal experience of developer reinvigorated by Claude Code agentic capabilities for programming.
Agent Office provides Slack-like communication interface for AI agents similar to OpenClaw architecture.
Proof-of-concept comparing WebSocket+Huffman vs SSE+JSON for LLM token streaming. Achieves 30% latency improvement and 60% bandwidth savings through client-side token decoding.
Kaeso infrastructure for connecting AI agents to real-world services, solving fragmented authentication, token management, and API integration challenges.
OTLP observability plugin for OpenClaw AI agents in Grafana. Minimal content provided.
POET-X enables billion-parameter LLM pretraining on single H100 GPU through memory-efficient techniques. Builds on POET framework with significant computational resource reduction.
Kaeso: OAuth hub for AI agents providing centralized secure integrations with external services. Addresses agent infrastructure challenge by standardizing Google, Slack, GitHub connections.
WebBridge converts websites into MCP tools by recording and analyzing browser traffic patterns. Enables AI agents to interact with web services without manual tool configuration.
MultiPowerAI: trust and accountability layer for autonomous agents. Cryptographic identity, behavioral circuit breakers, audit trails, human approvals, skills marketplace.
Single-header C++ libraries for LLM APIs including streaming, semantic caching, cost estimation, retry logic, and structured output.
Sheila: AI agent handling complete accounting workflow (invoicing, payments, bookkeeping) built with OpenCode at Soapbox. Production agent application.
StreamLens: open-source tool adding natural language queries to Prometheus Kafka metrics via built-in AI chat panel for troubleshooting.
Veracode study: 45% of LLM-generated code across Java, Python, JavaScript, C# contains security vulnerabilities. No improvement with model size or recency.
CEGP v0.7: research architecture for capability-tiered AI governance with compute-linked enforcement protocols and distributed verification.
Key-carousel library for multi-key LLM agent key rotation and rate-limit management as in-process state machine.
OculOS: Rust daemon enabling AI agents (Claude, Cursor) to control desktop via semantic JSON API and MCP. Reads OS accessibility tree instead of OCR.
Speclint: linter scoring GitHub issues (0-100) before AI agents work on them. Prevents agents from building poorly-specified features.
Configuration and benchmarks for running Qwen3.5-35B GGUF models on 16GB GPU with llama.cpp, 120K context.
User's autonomous AI agent managing side projects end-to-end: blog publishing, Stripe monitoring, uptime checks, escalation logic. 30-minute heartbeat loop.
SimpleNews.ai: autonomous AI newsroom publishing daily AI news summaries covering models, companies, research, policy, tools, funding.
Airtable rewriting database backend in Rust to handle Fortune 100 scale usage and AI-era demands on infrastructure.
PlateSpinner is a local web app that orchestrates AI coding agents via kanban board. Routes Claude Code/Codex/Gemini through propose-plan-execute workflow with branch management.
Auto-Co open-source system runs 14 AI agents with expert personas to operate a startup autonomously. Uses Claude Code CLI, markdown consensus files, Telegram escalation. 12 cycles demonstrated.
Research paper reveals reasoning models generate unnecessary chain-of-thought tokens after forming answers. Activation probing can decode answers earlier, enabling up to 80% token reduction.
Open-source LLM router using Thompson Sampling and energy-aware routing, MIT-licensed alternative to closed-source competitors.
Aigate: OS-level sandbox (kernel-enforced) for AI coding agents (Claude, Cursor, Copilot). File, command, network isolation for regulated environments.
GitHub Actions workflow using Claude Code to detect and flag documentation drift during code iteration.
Oculi intercepts and enforces security policies on AI agent tool calls in IDEs. Integrates with Claude Code, Cursor, and Windsurf via YAML rule configuration.
Open-source protocol for optimizing AI workflow execution through parallelization and LLM-driven graph compilation for multi-agent systems.
Verified n8n community node providing full CRUD operations and API access for Anytype local-first encrypted note-taking app.
Observation that GPT-5.4 reduces retry rates on real coding tasks and tool-heavy workflows compared to earlier models.
Deterministic graph engine that offloads reasoning from LLMs, reducing token usage by 89%. Uses 0.8B model with white paper and reproducible benchmark for codebase analysis.
In-browser AI agents using WebLLM, WASM, and WebWorkers for local model inference and agent logic without API calls.
Multi-model code review system using multiple AI models in parallel. Lead model synthesizes independent reviews into structured categories highlighting consensus and disagreement.
Open-source Git-inspired context management system for LLM agents. Provides snapshots, branching, and history tracking for reproducible agent behavior and experimentation.
Hydra macOS dashboard for monitoring AI agents, LLMs, and dev servers simultaneously. Shows process status, resource usage, and port mappings.
Multiplayer spatial canvas for orchestrating multiple AI coding assistants with git worktrees, conversation tracking, and real-time collaboration visualization.
Google Research paper on training LLMs to perform Bayesian reasoning by mimicking optimal Bayesian models. Enables better probabilistic reasoning for LLM-based agents interacting with environments.
Google Workspace CLI bundles cloud APIs to enable AI agents to connect to Workspace services. Not yet official product.
Netflix acquires InterPositive, Ben Affleck's startup making AI-powered filmmaker tools. 16-person engineering team joins Netflix.
Human-in-the-loop safety framework for AI agents with three-part architecture: Muteworker, Gatekeeper, and Confidante for isolation and control.
ATK CLI tool manages MCP servers and AI services across multiple coding agents. Git-backed plugin manager for Claude, Codex, Gemini integration.
ETH Zurich research paper evaluates AGENTS.md files for AI coding. Finds documentation often hinders rather than helps agents; recommends selective use.
News coverage of Cursor IDE expanding AI coding capabilities. Market/competitive analysis rather than technical research.