Stop building your AI stack from 40 open tabs.

lexmount 16.4k · today

Rust headless browser for AI agents on Servo/Stylo components: fetch pages as Markdown or a semantic tree, automate via CDP, WebDriver or BiDi — about 13% of Chrome's memory per task.

raphaelmansuy 100 · 7 days ago

Markdown to knowledge graph with an ontology you write: a small local decision model answers closed yes/no questions per candidate, and your cutoff keeps, queues for review or drops each link.

datalab-to 1.2k · 6 days ago

Datalab's 9B vision model that extracts JSON matching your schema from PDFs and images in one pass — schema-constrained decoding, multi-page values, vLLM or HF inference, Schema Studio.

thu-nics 718 · 2 days ago

Cache-to-Cache (ICLR'26): LLMs talk by projecting and fusing each other's KV-caches instead of generating text — higher accuracy than text relay at ~2x lower latency. Fuser weights on HF.

NVIDIA 19.9k · yesterday

NVIDIA's security scanner for agent skills: 71 patterns across 17 categories — prompt injection, exfiltration, supply chain, MCP tool poisoning — static plus optional LLM pass, SARIF output.

FalkorDB 9k · today

Low-latency, multi-tenant property graph database that runs queries as sparse-matrix linear algebra (GraphBLAS); OpenCypher over the Redis protocol, built for GraphRAG and agent memory.

strands-agents 8.8k · yesterday

Strands Agents: open-source Python and TypeScript SDK for building and running agent harnesses in your process — model-agnostic, with lifecycle controls, hooks, MCP, multi-agent and tracing.

strands-labs 579 · 2 days ago

Strands' small decision model (2B, Qwen3.5-based): pick between options, answer yes/no or rate on a scale with calibrated confidence. CLI and /v1/systemone server on CUDA, MPS, MLX or CPU.

walkinglabs 19.7k · 10 days ago

Project-based course on harness engineering for coding agents: 14 lectures, 8 projects on tools, state, verification and loops, plus breakdowns of Claude Code, Codex and Pi. 15 languages.

apache 2.4k · today

Apache Ossie (incubating, formerly Open Semantic Interchange): a vendor-neutral JSON/YAML spec for exchanging semantic models across BI, analytics and AI tools, with reference converters.

tinyhumansai 41.8k · today

Rust agent harness built for fleets: desktop app, terminal or embeddable library, with built-in memory, browser/desktop control, 119 OAuth apps and 14 channels — 500 agents on a $10 VPS.

microsoft 1.4k · yesterday

Microsoft's Go CLI for agent-skill quality: scaffold eval suites from SKILL.md, run them through Copilot, grade, compare models, replay snapshots and check spec and token budgets.

ag-ui-protocol 16.5k · yesterday

AG-UI, the Agent-User Interaction Protocol: ~16 standard event types that stream agent runs into frontends — chat, shared state, generative UI, frontend tools, human-in-the-loop.

petergyang 1.4k · 24 days ago

Skill + local tool to review HTML and Markdown like a Google Doc: edit text, drag blocks, comment on elements, then send every change to your coding agent in one batch.

angel291592 849 · today

Intent compiler skill for coding agents: probes your repo, asks only what no file answers, halts on ambiguity, and emits a typed IntentSpec it later checks the delivered work against.

Shubhamsaboo 141.3k · 11 days ago

141k-star collection of 100+ runnable AI agent, agent-skill and RAG app templates — clone one, pip install, run — across Claude, Gemini, GPT and open models. Apache-2.0.

google-research 1.5k · 18 days ago

Google Research's Regularized Recursive Self-Improvement: evolves an agent harness (prompts, tools, memory, sub-agents) with constraints on the search so gains hold out of distribution.

dzhng 2.6k · 9 days ago

Find code by asking what it does: `jg` uses the Jev decision model to judge folders, files and declarations, returning relevant files plus verbatim excerpts for coding agents. CLI + skill.

NVIDIA 15.7k · yesterday

NVIDIA's safe runtime for autonomous agents: kernel-enforced sandboxes with policy on every file, syscall and connection; credentials injected only for approved endpoints.

microsoft 18.2k · 5 days ago

Microsoft's text-space optimizer that trains a frozen agent's skill document like weights — rollouts, bounded edits, validation-gated updates — and ships a compact best_skill.md.

OpenBMB 386 · yesterday

OpenBMB's RL training framework where every role — inference, training, rollout — is an independent service talking over one TransferQueue data plane. Built on SGLang and torchtitan.

vllm-project 6.1k · yesterday

vLLM's programmable decision layer: one endpoint for your agent harness that selects or combines models per call by policy — quality, latency, cost, location — across local and cloud backends.

yetone 8.6k · yesterday

Pick any model for any coding agent from one menu-bar app: a local gateway translating OpenAI, Anthropic and Gemini APIs, with subscription sign-ins and quota failover across accounts.

spinabot 11.3k · 2 days ago

Self-hosted crew of AI agents on an org chart sharing one long-term memory: delegation, mid-task model switching, 1,000+ app connectors, reachable from terminal, WhatsApp, Telegram or Slack.

block 35.8k · yesterday

Block's self-hostable workspace where humans and agents share rooms on a Nostr relay — chat, canvases, workflows, patches and reviews as signed events; agents join with their own keys.

tigerless-labs 11k · 2 days ago

Self-learning skill layer for Claude Code: distills skills from your real sessions, merges same-scenario ones, updates them in use and prunes the unused — touching only skills it wrote.

aru-labs 143 · 18 days ago

Lossless long-term memory for a personal AI: verbatim JSONL logs, time-first search over SQLite FTS5 (sqlite-vec as last resort) and a tiny 'where are we now' index injected every turn.

coldteadotai 572 · yesterday

Hooks Claude Code, Codex, OpenCode and Pi so every edit is checked against the rules in your AGENTS.md/CLAUDE.md by a decision model (Jev) — ~300 ms per check, violations fixed in-turn.

dream-num 22.5k · yesterday

Office SDK for embedding spreadsheets, docs, slides, bases and PDFs in your own product — canvas rendering, formula engine, plugins and one Facade API across browser and Node.js.

Contrastive-LM 3k · 6 days ago

Contrastive Language Models: CLM-8B, an open System-1 decision model scoring states against actions — typed choice/score/yes-no answers on a TypeSafe-compatible API, up to 9x faster than Jev.

volotat 1.2k · yesterday

Continual-learning byte-level LM trained from scratch on one 8 GB GPU: weights page in from disk, capacity grows and prunes itself while it keeps reading a single data stream.

trycua 29k · 3 days ago

Computer-use infrastructure for agents: sandboxed Linux, macOS and Windows desktops locally or in the cloud, Cua Driver for native apps (CLI/MCP/SDK), Lume VMs, CUA-S1 models, Cua Bench.

run-llama 12.8k · 6 days ago

LlamaIndex's local document parser in Rust: PDFium text with bounding boxes at ~2-5 ms/page, selective Tesseract or HTTP OCR, Markdown/JSON output, screenshots; Python, Node, WASM.

google 13.3k · 14 days ago

Google's declarative runtime for agent workloads on Kubernetes: Task, Workspace and Model manifests run each agent sandboxed on Agent Substrate, with suspend/resume and ax ssh.

nokia-applied-research 1.1k · 4 days ago

Nokia research: turn any open LLM into a Jev-style decision model. Typed choice, yes/no or score from one prefill, de-biased with no labels or a fitted head, served on vLLM.

NandhaKishorM 31.6k · 4 days ago

Non-autoregressive decision engine: typed choice, score and yes/no answers over text in one forward pass (~33 ms), 100+ languages, calibrated probabilities, a router picking the checkpoint.