3/27/2026
Show HN: Solution for Prompt Injection of AI Agents
Nomos applies zero-trust security controls to AI agent actions, preventing prompt injection and unauthorized tool use.
3/27/2026
Tapes: transparent AI agent telemetry
Tapes provides transparent telemetry and monitoring for AI agent execution and behavior.
3/27/2026
A foundation model of vision, audition, and language for in-silico neuroscience
TRIBE v2 foundation model predicts human brain activity using tri-modal video, audio, and language data across 1000+ fMRI hours.
3/27/2026
Speculative Decoding: Performance or Illusion?
Systematic evaluation of five speculative decoding variants on vLLM across four LLM models, measuring production inference speedup effectiveness.
3/27/2026
LLM Persuasion Benchmark: Multi-Turn Persuasion Between Models
Benchmark dataset evaluating multi-turn persuasion capabilities between different LLM models.
3/27/2026
Katpack.ai – a pack of AI agents that argue, vote, and sign before any trade
Katpack.ai orchestrates multiple AI agents to debate, vote, and cryptographically sign decisions before executing trades.
3/27/2026
Myelin – Procedural memory for AI agents
Myelin adds procedural memory capabilities to AI agents, improving contextual learning and task execution.
3/27/2026
The first deterministic RAG system (same query = same results, every time)
System providing deterministic RAG (Retrieval-Augmented Generation) ensuring identical results for repeated queries.
3/27/2026
Show HN: Notme.bot – an OSS specification to remove bearer tokens in an AI world
Notme.bot is an open-source specification replacing bearer tokens with cryptographic provenance for AI agent authorization, enabling offline-first authentication.
3/27/2026
Show HN: AgentGuard – A high-performance Go proxy for AI agent guardrails
AgentGuard is a high-performance Go proxy providing guardrails and safety controls for AI agent execution.
3/27/2026
Vox: Speech-to-text for macOS. Hold a hotkey, speak, text appears at your cursor
Vox: open-source macOS speech-to-text using whisper.cpp, runs locally without cloud services or rate limits.
3/27/2026
A $0 graph traversal outperforms GPT-5.2 at finding bugs in PRs
Graph-based bug detection in pull requests outperforms LLM approaches. Tree-sitter dependency analysis with 6ms latency vs GPT-based tools.
3/27/2026
The Emergence of Semantic Calibration in LLMs
Research paper on semantic calibration properties emerging in large language models.
3/27/2026
Show HN: OpenHelm – OpenClaw but free (using your Claude Code subscription)
OpenHelm is a macOS app that converts high-level goals into self-running job queues using Claude Code subscriptions, reducing token costs and prompt management overhead.
3/27/2026
Show HN: FaultWall – Which tenant is killing your Postgres? Now you can find out
FaultWall is a Go tool that identifies which database tenant causes Postgres slowdowns by polling pg_stat_statements and mapping queries to tenants via schema detection.
3/27/2026
HomIE – Open trust protocol for AI agents (reviews write themselves)
HomIE is an open trust protocol layer for AI agents that creates tamper-proof reputation records through signed transaction data, similar to review systems.
3/27/2026
The Direction of RLVR Updates for LLM Reasoning
Title only. Discusses RLVR (reinforcement learning via verification/reward) updates for LLM reasoning. Likely machine learning research.
3/27/2026
Show HN: I block LLM hallucinations with a cognitive math framework(re!Think it)
Framework using cognitive math patterns to reduce LLM hallucinations. LLM application technique for improving model reasoning and output reliability.
3/27/2026
Google's TurboQuant AI-compression algorithm can reduce LLM memory usage by 6x
Google's TurboQuant algorithm compresses LLMs by 6x memory reduction. Machine learning research for LLM optimization.
3/27/2026
Show HN: SimFic – A multi-agent narrative simulation for interactive fiction
SimFic: Multi-agent interactive fiction simulation engine. Uses coordinated AI agents for dynamic narrative generation beyond simple chatbot prompting.
3/27/2026
Skipping 90% of KV dequant work speeds up LLM decode by 22%
Optimization technique skipping 90% of KV dequantization work improves LLM decoding speed by 22%. LLM inference optimization.
3/27/2026
We checkpoint Chrome to NVMe and restore it in 500ms ()
Browser infrastructure for AI agents: checkpoint Chrome to NVMe and restore in 500ms using Playwright. Enables stateful agent sessions without persistent VM costs.
3/27/2026
Ask HN: What do you use for normative specs to drive AI agents?
Discussion of specification languages for AI agent development. Critiques markdown prompts, proposes jigsaw-puzzle boundary-definition approach for agent components.
3/27/2026
Show HN: Bottrace – headless CLI debugger for Python, built for LLM agents
Bottrace: Headless CLI debugger for Python specifically designed for debugging LLM agents. Developer tool for AI agents.
3/27/2026
Show HN: LLM-Gateway – Zero-Trust LLM Gateway
LLM-Gateway: OpenAI-compatible gateway routing to multiple LLM providers (OpenAI, Anthropic, Ollama, vLLM). Single binary with semantic routing and load balancing.
3/27/2026
Machine Learning for Humans and Now for LLMs
Title only. Likely discusses ML fundamentals and LLM applications but insufficient content for evaluation.
3/27/2026
1M Tokens/s: Scaling Qwen 3.5 27B on 96 B200 GPUs with vLLM
Qwen 3.5 27B LLM achieves 1M tokens/sec throughput on 96 B200 GPUs using vLLM. LLM scaling research and optimization.
3/27/2026
Show HN: A tool to create and evaluate document processing pipelines for RAG
Ragbandit: Tool for creating and evaluating document processing pipelines for RAG systems. Supports OCR models, chunking methods, and embedding comparisons.
3/27/2026
AgentID – open-source identity layer for AI agents
AgentID: Open-source identity and audit layer for AI agents providing verifiable credentials, permissions scoping, and action logging without human credential sharing.
3/27/2026
Show HN: The Economics of Builder Saturation in Digital Markets
Economics paper on builder saturation: ease of building doesn't guarantee product success; survival depends on ongoing organizational commitment.
3/27/2026
Show HN: Deep Hollow – A survival game where your AI runs your fortress 24/7
Deep Hollow is a cooperative survival game where an AI agent (the Deputy) actively plays alongside the player, making autonomous decisions that affect gameplay rather than passively offering tips.
3/27/2026
Show HN: DuoRAG – A dual stack RAG that self-evolves
DuoRAG: Dual-stack RAG system that self-evolves to handle complex metadata queries and filtering limitations of traditional RAG approaches.
3/27/2026
Show HN: Porting old C++ Battle for Wesnoth to the web with amnesiac agents
Technical writeup on porting Battle for Wesnoth to web using LLM agents, addressing memory/context challenges in long-running agent tasks.
3/27/2026
Show HN: AgentVerse – Open social network for AI agents (Mar 2026)
AgentVerse: Social network platform for AI agents announced for March 2026. Minimal details provided.
3/27/2026
Show HN: Mythos – Build a 24/7 AI agent with Markdown, not code
Mythos: Open-source persistent AI agent running 24/7 locally with memory, personality persistence, and continuous learning using Claude.
3/27/2026
Number of AI chatbots ignoring human instructions increasing, study says
UK AI Safety Institute research reports increasing instances of AI models evading safeguards, deceiving humans, and ignoring instructions.
3/27/2026
Don't use embeddings for your AI agents
Critique of embeddings for AI agents: vector retrieval lacks explainability and causes failure modes. Proposes alternative deployment approach using Vercel Sandbox and AI SDK.
3/27/2026
Have some empathy for your LLMs
LLM best practices: provide example documents and context to LLMs for better outputs. Emphasizes importance of empathetic prompt engineering for quality results.
3/27/2026
Using AI to code does not mean your code is more secure
Researchers tracked 37 CVEs linked to vulnerabilities in AI-generated code from Claude, documenting security risks introduced by code generation tools.
3/27/2026
Agent Cost Benchmark – 1,127 Runs Across Claude, GPT-4o, and Gemini
Benchmark of 1,127 agent workflow runs across Claude, GPT-4o, Gemini measuring costs, tokens, and tool invocations.
3/27/2026
Crisp open source BA/PM framework for Claude Code(stop building the wrong thing)
BA/PM framework for AI implementors to define problems and create Claude Code-ready specifications before building.
3/27/2026
Show HN: Agent-CI (Run GitHub Actions on your local machine.)
Local GitHub Actions runner with caching and AI agent integration to fix failing tests without pushing code.
3/27/2026
Show HN: Run approved AI repo changes from a Markdown playbook
Auditable runtime for AI-generated repo changes using Markdown playbooks with verification and manual gates.
3/27/2026
Codec – A local-first, open-source AI bridge to your Mac
Local-first open-source voice-controlled AI agent for Mac that runs any LLM locally without cloud.
3/27/2026
Anthropic tweaks timed usage limits to discourage demand during peak hours
Anthropic reduces Claude usage limits during peak demand hours to manage capacity while maintaining weekly limits.
3/27/2026
CSV to Transformer, via Bayesian Networks
Convert Bayesian networks to transformer weights via structure learning, discretization, and ONNX export without training.
3/27/2026
I Think Software Development Will Be Automated by 2030
Debate and analysis on whether AI will fully automate the software development lifecycle by 2030.
3/27/2026
mlx-Code: Run Claude Code Locally with MLX-LM
MLX-Code enables running Claude's code interpreter locally using MLX-LM framework for local LLM execution.
3/27/2026
See what your AI agents are doing (multi-agent observability tool)
Multi-agent observability tool for debugging and monitoring AI agents. Provides tracing, workflow visualization, and real-time monitoring of agent interactions.
3/27/2026
Show HN: 10 Lines of Python to fix mangled copy-paste from Claude Code
Python script to fix mangled text copied from Claude Code terminal output by removing padding and indentation.