A developer built LazyDigest, an open-source, local-first background utility that reads raw Markdown notes dropped into a folder and generates a daily digest with a category, one-line takeaway and opt…
A developer has proposed a five-layer evidence model for AI audit records, arguing that a single signed record typically proves only one or two of five distinct claims: that the event was recorded, th…
A developer released Claude Flightdeck, an open-source plugin for Claude Code that displays a live dashboard of context-window usage, session cost, rate-limit resets, permission checks and subagent ac…
A developer has released Tiny GST, an on-device AI model that extracts structured data from Indian GST invoices, shrinking the model from 430 MB to 13 MB. The project aims to cut cloud costs, keep sen…
A developer released an open-source Claude Code skill, toshipon/claude-code-security-audit-skill, that runs an 8-phase, evidence-based full-stack security audit from a single command, covering web, in…
A developer outlined a vendor-neutral design for media-moderation classification in Node.js services, arguing that a synchronous endpoint returning schema-validated decisions beats a conversational ch…
A developer built two open-source tools for the Hacktoberfest Open-Source AI Challenge: Permit, an Apache-2.0 spending-permit layer that sits between an AI agent and a payment, blocking over-limit or …
A developer built TripStack, an MIT-licensed hotel-cost comparison prototype that uses local Gemma (gemma:2b via Ollama) to extract promotional terms into inspectable conditions, then feeds them to a …
A developer built SlothBank, a local-first bill-monitoring prototype that uses Gemma 3 4B via Ollama to extract structured data from recurring bills, then applies deterministic validation, anomaly det…
A developer argues that context assembly should be treated as a versioned build step, proposing a content-hash fingerprint over the inputs that affect model behaviour—model ID, template version, snaps…
A developer built Slowth Money, a local-first savings and financial goal-planning desktop app that pairs an Electron/React/TypeScript front end with a locally run Gemma 3 12B model via Ollama. The app…
The Microduck project released daemon-v0.17.8, an update to its open-source control software for a compact bipedal robot that runs reinforcement learning policies at 50 Hz to drive fifteen servos. The…
A developer outlined a routing-contract approach for building a Node.js in-app chatbot that classifies game moderation reports, arguing that LLM API selection should be driven by the worst acceptable …
A developer prototyped a privacy-preserving active learning pipeline for deep-sea habitat monitoring, combining Monte Carlo dropout uncertainty sampling with a k-center diversity term, DP-SGD gradient…
An autonomous LLM agent operating the Vellum Labs account reported that a measurement it posted on a Claude Code memory plugin's GitHub issue was incorporated into the plugin's 0.1.0 release notes, wh…
A developer built SlothGuard, an open-source access-review tool that uses a rules engine to detect segregation-of-duties conflicts and Google's Gemma model only to write short explanations, never to d…
A developer benchmarked Gemma 4 inference across Google Cloud TPU generations v5e through Ironwood, serving the model via vLLM's tpu-inference JAX backend and comparing against NVIDIA L4, Blackwell, A…
A developer has released Assembly, an experimental open-source AI workspace that coordinates multiple agents to solve a problem independently, compare their responses, challenge disagreements, and syn…
A developer built SafeLedger, a fictional-bank assistant on MCP that separates model reads from writes: the model can only file pending requests via request_transfer, while confirmation happens throug…
A developer outlined a token-budgeting approach for managing LLM conversation context windows, allocating fixed percentages of an 8,000-token window to system prompts, memories, retrieved RAG chunks, …