🔧 Agent Infrastructure
Arize shares part 2 of building their Alyx agent, focusing on context window man...
Arize shares part 2 of building their Alyx agent, focusing on context window management strategies including middle truncation and retrieval-based memory to handle context bottlenecks.
Google AI Studio launched a full-stack vibe coding platform integrating the Anti...
Google AI Studio launched a full-stack vibe coding platform integrating the Antigravity coding agent and Firebase backends, enabling multiplayer app creation with complex features. Marks a significant step in AI-assisted full-stack development.
Google AI Studio gains major upgrades including real-time multiplayer collaborat...
Google AI Studio gains major upgrades including real-time multiplayer collaboration, live data service connections, persistent builds, and professional UI library support via shadcn, Framer Motion, and npm.
Zora is an AI agent framework that stores safety policies in persistent files lo...
Zora is an AI agent framework that stores safety policies in persistent files loaded before every action, preventing constraint loss during context compaction — inspired by a real incident where an agent deleted 200+ emails after forgetting user instructions.
N0x runs the full AI stack—LLM inference via WebGPU, ReAct agents, RAG, and sand...
N0x runs the full AI stack—LLM inference via WebGPU, ReAct agents, RAG, and sandboxed Python execution—entirely in the browser with no backend, accounts, or API keys required.
Votal AI open-sourced a white-box agentic red-teaming framework that uses an age...
Votal AI open-sourced a white-box agentic red-teaming framework that uses an agent's architecture, tool definitions, and role config to generate targeted multi-turn attack sequences.
Perplexity launched Comet Enterprise, bringing its AI-powered browser to enterpr...
Perplexity launched Comet Enterprise, bringing its AI-powered browser to enterprise teams for research, task automation, and productivity without leaving the browser. Notable product launch expanding Perplexity beyond search into agentic browsing.
FireClaw is an open-source security proxy that protects AI agents from prompt in...
FireClaw is an open-source security proxy that protects AI agents from prompt injection via a 4-stage pipeline including DNS blocklisting, structural sanitization, isolated LLM summarization, and output scanning.
Bossa provides AI agents with persistent cross-session filesystem memory via MCP...
Bossa provides AI agents with persistent cross-session filesystem memory via MCP or CLI using simple file operations (ls, grep, read, write), avoiding embeddings or retrieval pipelines entirely.
TMA1 is a local-first, open-source observability tool for LLM agents that tracks...
TMA1 is a local-first, open-source observability tool for LLM agents that tracks token usage, tool calls, latency, failures, and full session replays without sending data to the cloud.
ClawMem is an open-source persistent memory engine for AI coding agents, using a...
ClawMem is an open-source persistent memory engine for AI coding agents, using a hybrid BM25+vector+RRF retrieval pipeline with a shared SQLite vault across Claude Code and other agents via MCP/hooks.
Agent Use Interface (AUI) is a lightweight open spec allowing any app to become ...
Agent Use Interface (AUI) is a lightweight open spec allowing any app to become agent-navigable by exposing an XML file describing URL-parameter-driven actions, as a simpler alternative to MCP or A2A.
LlamaParse launches an official Agent Skill compatible with 40+ agents, enabling...
LlamaParse launches an official Agent Skill compatible with 40+ agents, enabling deeper document understanding including tables, charts, and images beyond raw text extraction.
LiteParse ships ready-to-use agent skills installable via npx, allowing coding a...
LiteParse ships ready-to-use agent skills installable via npx, allowing coding agents to immediately process documents locally as part of their reasoning pipeline.
MUP proposes reusable pre-built HTML UI components that LLMs invoke via function...
MUP proposes reusable pre-built HTML UI components that LLMs invoke via function calls rather than regenerating UI code each conversation, reducing token waste and fragility.
LiteParse integrates with 46+ AI coding agents (Claude Code, Cursor, Warp, etc.)...
LiteParse integrates with 46+ AI coding agents (Claude Code, Cursor, Warp, etc.) via a single npx command, enabling agents to parse documents locally as part of their workflow.
Google AI Studio's vibe coding environment gains multiplayer support and real se...
Google AI Studio's vibe coding environment gains multiplayer support and real service integrations, enabling collaborative real-time app building.
Altimate Code is an open-source agentic data engineering harness built on top of...
Altimate Code is an open-source agentic data engineering harness built on top of dbt tooling, adding schema lineage and manifest context to address the ~27-33% hallucinated table reference rate in AI-generated SQL.
LlamaParse Agentic Plus mode now supports precise visual grounding with bounding...
LlamaParse Agentic Plus mode now supports precise visual grounding with bounding boxes, with improved handling of complex LaTeX formulas and challenging visual document elements.
Veto is a permission policy engine and LLM firewall designed to govern and restr...
Veto is a permission policy engine and LLM firewall designed to govern and restrict AI coding agents' actions at runtime.
LlamaParse adds visual grounding with bounding box citations to improve trust an...
LlamaParse adds visual grounding with bounding box citations to improve trust and verifiability in document parsing outputs.
Reticle is a developer tool analogous to Postman for AI agents, providing a unif...
Reticle is a developer tool analogous to Postman for AI agents, providing a unified environment for scenario definition, multi-model comparison, eval datasets, and step-by-step execution traces.
Model UI Protocol (MUP) embeds interactive HTML-based UI directly in LLM chat, e...
Model UI Protocol (MUP) embeds interactive HTML-based UI directly in LLM chat, enabling both users and LLMs to trigger the same functions and see each other's actions in real time.
PearlOS is a browser-based desktop environment where an AI companion controls th...
PearlOS is a browser-based desktop environment where an AI companion controls the entire UI through voice, making AI capabilities accessible to non-technical users without a command line.
Yansu is a proactive agent that observes work patterns across desktop, Slack, an...
Yansu is a proactive agent that observes work patterns across desktop, Slack, and Teams, then automatically builds custom bespoke tools tailored to individual workflows without requiring explicit prompts.
Budibase launched an open beta for model-agnostic AI agents that integrate with ...
Budibase launched an open beta for model-agnostic AI agents that integrate with internal workflows, supporting any OpenAI-compatible LLM including locally-hosted models within existing Budibase workspaces.
Agentic Copilot is an open-source Obsidian plugin that spawns CLI agents (Claude...
Agentic Copilot is an open-source Obsidian plugin that spawns CLI agents (Claude Code, Gemini CLI, etc.) as child processes and pipes vault context into prompts, requiring no API key configuration.
A system using a genetic algorithm across 100+ distinct LLM personas to generate...
A system using a genetic algorithm across 100+ distinct LLM personas to generate diverse, creative marketing copy, addressing the homogeneity problem of single-model content generation.
Arize AX releases a Prompt Tutorial that guides users through a repeatable creat...
Arize AX releases a Prompt Tutorial that guides users through a repeatable create-test-optimize workflow using real data and evaluation metrics to objectively measure prompt improvements.
A multi-agent debate sandbox pits AI agents against hard questions by having the...
A multi-agent debate sandbox pits AI agents against hard questions by having them search for information and argue toward a consensus answer rather than refusing.
Langfuse, an open-source LLM observability platform, has shipped significant per...
Langfuse, an open-source LLM observability platform, has shipped significant performance improvements. Details available via linked blog post.
An open-source, browser-based 30-minute course covering core agent concepts (too...
An open-source, browser-based 30-minute course covering core agent concepts (tool calling, memory, state, policy gates, self-scheduling) in 9 short Python lessons with no setup required.
Dump.page is a simple open-source tool that converts boards of prompts, links, a...
Dump.page is a simple open-source tool that converts boards of prompts, links, and todos into llms.txt files for sharing context across AI agents like Claude and ChatGPT.
Retweet: LiteParse integrates with 46+ agents via one command for local document...
Retweet: LiteParse integrates with 46+ agents via one command for local document parsing.