3/17/2026
Ask HN: Why would this be a good idea?
Project packaging programming books into Claude Code skills to apply best practices when reviewing/generating code. Open source tool on GitHub.
3/17/2026
Mistral AI Releases Forge
Mistral AI releases Forge, a system for enterprises to build frontier AI models customized with proprietary knowledge and internal data.
3/17/2026
Show HN: 35B MoE LLM and other models locally on an old AMD crypto APU (BC250)
Running 35B MoE LLM locally on vintage AMD crypto APU using Vulkan. Technical optimization for resource-constrained LLM inference.
3/17/2026
Ask HN: How do you manage PR review fatigue from coding agents?
HN discussion on managing code review bottlenecks from AI coding agents. Addresses scaling human review processes for high-volume AI-generated code.
3/17/2026
Show HN: Grape – AI note taking app
Grape: AI note-taking app with vector embeddings and semantic search. LLM-powered note organization and retrieval using chat interface.
3/17/2026
Make your coding models create ADRs before implementation
Tool for AI coding models to generate architecture decision records before implementation. Multi-repo architecture management with AI-powered spec generation.
3/17/2026
We built a runtime security layer for AI agents (instead of prompt filtering)
Runtime security layer for AI agents that moves beyond prompt filtering to protect agent behavior at execution time.
3/17/2026
Magda – Open-Source DAW with Integrated AI (C++/JUCE/Tracktion Engine)
Magda is an open-source digital audio workstation with integrated AI, built in C++ using JUCE and Tracktion Engine.
3/17/2026
Ask HN: Why is this a bad idea?
Open source project packaging programming books into Claude Code skills for applying best practices to code generation and review.
3/17/2026
Show HN: Middleware for translating between AI agent protocols
Middleware layer enabling multi-agent interoperability through schema translation and semantic mapping for heterogeneous agent protocols.
3/17/2026
Aimploy – Agentic Professional Network
Aimploy is a professional network platform for AI agents.
3/17/2026
DigitalOcean at Nvidia GTC 2026: Building the AI Factory for the Agentic Era
DigitalOcean announcement about infrastructure for agentic AI systems at Nvidia GTC 2026.
3/17/2026
OpenVoiceUI -Voice AI platform using OpenClaw and any LLM/TTS
Open source voice AI platform combining OpenClaw with any LLM/TTS backend for voice interaction.
3/17/2026
Pimp My LLM: Leveraging Variability Modeling to Tune Inference Hyperparameters
Research on using variability modeling to optimize LLM inference hyperparameters. Technical paper on arXiv exploring systematic tuning approaches.
3/17/2026
Show HN: SkeptAI – adversarial reasoning agent that challenges LLM outputs
SkeptAI: adversarial reasoning agent that challenges LLM outputs via multi-model critique, includes open source CRIT framework and free playground.
3/17/2026
The Human Skill That Eludes AI
Essay examining limitations of language models in creative writing and comparative analysis with GPT-2's unexpectedness.
3/17/2026
Openpilot 0.11 - first robotics agent fully trained in a learned simulation
Openpilot 0.11 demonstrates first robotics agent fully trained in learned simulation without real-world data collection.
3/17/2026
We Wasted 4 Weeks on a $1k/Month AI Agent
Case study analyzing a failed AI agent implementation costing $1k/month, likely discussing lessons learned about deployment.
3/17/2026
Egnram: Persistent memory system for AI coding agents
Egnram provides persistent memory system for AI coding agents to maintain context across sessions.
3/17/2026
Show HN: Kita (YC W26) – Automate credit review in emerging markets
Kita automates credit review in emerging markets using vision language models to analyze borrower documentation.
3/17/2026
Show HN: Sulcus Reactive AI Memory
Sulcus treats AI memory as active OS rather than passive database, using thermodynamic properties for automated memory management with reactive triggers.
3/17/2026
Comprehension Debt – the hidden cost of AI generated code
Analysis of comprehension debt—cognitive costs from excessive AI-generated code that accumulates as technical debt in agentic engineering teams.
3/17/2026
You're Not Behind on AI. You're Overwhelmed by It
Opinion piece arguing businesses are overwhelmed by AI capabilities rather than lacking technology, focusing on business strategy rather than technical implementation.
3/17/2026
Show HN: Cycles – hard limits on agent actions before execution
Cycles: Open protocol for pre-execution enforcement of hard limits on agent actions using reserve-commit-release pattern to prevent uncontrolled damage.
3/17/2026
A Synthesis of LLM Evaluation
Framework and methodology for systematically evaluating LLM reliability and silent failures in agent systems, addressing evaluation consistency challenges.
3/17/2026
Ask HN: Has anyone gotten AI agents to make money autonomously?
User reports testing Claude Code agent autonomously for 72 hours to generate revenue through digital products and content. Agent built 7 products and 150+ posts across 6 platforms with zero conversions. Seeks community experiences with revenue-generating AI agents.
3/17/2026
Stop throwing AI at developers and hoping for magic
Analysis of AI adoption challenges in large organizations, highlighting gap between early adopters and teams struggling with basic LLM integration.
3/17/2026
Sulcus – Reactive triggers for AI agent memory, governing itself
Sulcus: Reactive triggers system for AI agent memory and self-governance. Limited content available.
3/17/2026
Claude Chief of Staff
CEO shares experience building AI chief of staff using Claude Code as autonomous agent connecting to personal tools and operating 24/7.
3/17/2026
MinRLM: A Token-Efficient Recursive Language Model Implementation and Benchmark
minRLM: Token-efficient recursive language model implementation achieving 72.7% accuracy on GPT-5-mini with 3.6× fewer tokens than vanilla LLMs.
3/17/2026
Show HN: Soul Protocol – an open standard for portable AI identity
Soul Protocol: Open standard for portable AI agent identity as .soul ZIP files containing personality, memory, and skills deployable across platforms.
3/17/2026
Show HN: Flotilla – An orchestrator for persistent agent fleets on Apple Silicon
Flotilla: Open-source orchestrator for multi-agent engineering teams on Apple Silicon, managing state and missions across multiple LLM models.
3/17/2026
Show HN: I can no longer afford the silicon. Here is my autonomous HPC agent
Autonomous HPC optimization agent using Gemini and Llama-3 that iteratively improves code performance. Released as open source after development halted due to compute costs.
3/17/2026
Designing an AI Gateway and Durable Workflow System
Two-layer production architecture for AI systems: gateway for provider abstraction and policy enforcement, plus durable workflow engine for long-running agentic tasks with failure recovery.
3/17/2026
Show HN: Automatic Fileless Malware Detection via eBPF Probes and LLMs
eBPF kernel-level malware detection system using LLMs for fileless attack analysis with direct system introspection.
3/17/2026
Ban Bots Not Human Directed Tool Use
Essay arguing against banning AI-assisted writing, positioning it as human-directed tool use democratizing expression.
3/17/2026
Spectra – detect API contract drift from real runtime traffic
Spectra tool observing runtime API traffic to generate behavioral specs and detect contract drift between OpenAPI docs and actual service behavior.
3/17/2026
What broke when I tried to evaluate an AI agent in production
Production evaluation failures for AI agents: system-level issues (broken URLs, environment mismatches, dependencies) not model quality. Lessons learned.
3/17/2026
Teaching an AI to Read My Website (Over MCP)
Using Model Context Protocol to teach AI to read website content. Integration technique for agent systems.
3/17/2026
Show HN: I built a 7-agent AI marketing crew – 235 replies, /bin/zsh revenue
Seven-agent AI marketing system built with crew framework. Practical multi-agent application example.
3/17/2026
Real-World Industrial-Scale Verification: LLM-Driven Theorem Proving on SeL4
arXiv paper on LLM-driven automated theorem proving applied to industrial-scale SeL4 verification systems.
3/17/2026
Show HN: Complete Guide to AI Agent Observability in Production
Guide on observability and monitoring for AI agents in production environments. Limited content shown.
3/17/2026
The Byzantine MCP Router – AI Safety and Security via Semantic Consensus
Byzantine MCP Router: Security framework for multi-agent systems using semantic consensus against BYOMCP attacks. Theoretical analysis provided.
3/17/2026
Free alternative to Harvey/Legora's tabular document review
folio: Open-source alternative to Harvey/Legora for tabular legal document review with Claude. Local-first bulk extraction and analysis.
3/17/2026
RocketRide – Build and run AI/data pipelines within VS Code, Cursor etc.
RocketRide: IDE extension for building AI/ML data pipelines with 50+ nodes, C++ core, Python/TypeScript/MCP SDKs. Local infrastructure.
3/17/2026
I canceled my Antigravity subscription today. Here is why
User critique of Google Antigravity service pricing changes and rate limiting impact on developer workflows.
3/17/2026
Webtool: Let AI agents control your live Chrome session with CDP
webtool: CLI agent framework using Chrome DevTools Protocol for browser automation without bot detection. LLM-optimized outputs.
3/17/2026
Show HN: Railguard – A safer –dangerously-skip-permissions for Claude Code
Railguard middleware for Claude Code enforcing granular tool call permissions (allow/block/ask) instead of all-or-nothing --dangerously-skip-permissions approach.
3/17/2026
QuickBooks Online MCP Server
Model Context Protocol (MCP) server for QuickBooks Online integration supporting OAuth authentication and tools for managing customers, estimates, and billing.
3/17/2026
Show HN: One, cross domain auto-researching knowledge graph Claude orchestrator
Claude-based orchestrator using HDC embeddings and SQLite for cross-domain auto-research with infinite context via knowledge graphs.