Context.dev (also @getcontextdev; YC S26 per founder/company bios) markets itself as a unified web-context API for software and agents: scrape, crawl, search, brand/people enrichment, monitors, MCP, and SDKs — one API key.
GitHub description: A free, open-source native macOS app for Docker & Linux containers, an alternative to OrbStack and Docker Desktop. Universal for Intel and Apple silicon. Homepage field: augani.github.io/dory. README hero: Your complete local Linux workspace, built for Mac.
Landing hero: Connect any agent to everything. Tagline: The gateway to connect your agent to everything. Page: Executor is an MCP gateway. Anything that speaks MCP, like Claude Code, Cursor, or Codex, points at one endpoint and reaches every tool you connect.
Maple is a general observability platform for traces, logs, and metrics. It accepts OTLP data and uses ClickHouse-family storage and query paths. Its product surfaces include distributed traces, correlated logs, metrics and dashboards, service maps, error tracking, alerts, infrastructure context, browser session replay, CLI queries, and an MCP server.
PageIndex (MIT, Python) is VectifyAI’s vectorless, reasoning-based RAG engine. First-party pitch: build a hierarchical tree index per document, then let an LLM reason through that tree (human-like section navigation) instead of embedding chunks into a vector database.
Open-source application platform to build, deploy, operate, and scale apps on infrastructure you own (VPS, dedicated, cloud VM, homelab). Pitch: replace a pile of deployment tools + managed services with one OSS control plane. Interfaces claimed: web dashboard, native desktop app, CLI, REST API, MCP for AI agents.
bb is an MIT-licensed, self-hosted agentic IDE that coordinates installed coding-agent CLIs in persistent, steerable threads. Its first-class control surfaces are a desktop app, web app, CLI, and HTTP API. The project’s central differentiator is that much of its UI/workflow surface is implemented as plugins, so an agent can extend the IDE itself rather than only working inside a fixed product.
Box is a managed sandbox product from Ascii (legal entity Dedale AI, Corp.). It sells persistent Linux VMs for coding agents and agent platforms: SSH, Docker inside the VM, a dedicated IPv4 per machine, snapshot/stop/resume, fork-from-snapshot, optional 60fps desktop stream, and a CLI/API/box prompt path for Claude Code or Codex.
Ditto is an open-source deterministic website cloner from ion.design. Point it at a public URL and it returns a cloned implementation as componentized Next.js or Vite code (Tailwind styling), with marketing copy citing ~5 minutes end-to-end.
MarkItDown (microsoft/markitdown) is a lightweight Python utility that converts many file types to Markdown for LLMs and text-analysis pipelines. README positions it near textract, but focused on preserving structure (headings, lists, tables, links) as Markdown for models rather than high-fidelity human publishing.
README: a CLI for exploring OpenTelemetry traces, metrics, and logs locally. Go custom Collector distribution, DuckDB backend, Svelte UI. Receives OTLP on localhost:4317 (gRPC) and localhost:4318 (HTTP). UI at localhost:8000. In-memory by default; --db persists DuckDB.
Sitegeist is an AI assistant that lives in a Chromium side panel, plus a bridge so coding agents can use the same browser. README framing: collaboration over “autonomy theater” — you guide, it executes. Automates repetitive web tasks, extraction, navigation, forms, research → docs/spreadsheets; BYO API key or subscription login (Anthropic, OpenAI/ChatGPT, GitHub Copilot, Gemini, etc.); claims local-only / no tracking.
Agentation is a visual feedback layer for agents working on web UIs. Humans click elements, annotate, optionally use layout mode (rearrange/resize), then copy structured markdown (or push via MCP/webhooks) so the agent gets element paths, CSS selectors, positions, and related context instead of vague prose.
README: "A Claude skill that explains an idea by building it as a real, working object." Ask for a figure that explains a technical idea; Anatomy invents a physical machine whose working is the concept, then builds it from dozens of shaded isometric parts in a framed card with a title, hint and live readout.
Concentrate is a managed LLM gateway: one API base URL (https://api.concentrate.ai/v1) for hundreds of models across all major providers (OpenAI, Anthropic, Gemini, DeepSeek, etc.). You pick the model and provider; Concentrate runs the request, applies your fallbacks and limits, and surfaces spend/logs.
fx is a Vercel Labs coding-agent harness and CLI written in Zig. Official copy positions it as a tiny, open, native agent originally used internally, now open sourced for research and embedding in larger systems. The durable object is the product at fx.sh / vercel-labs/fx, not the tweet.
gws is a single Google Workspace CLI. It dynamically reads Google's Discovery Service at runtime and builds its command surface from the available APIs instead of shipping a static command list. The README positions it as built for both humans and AI agents: every response is structured JSON, and the repo includes 40+ agent skills.
Hark turns HTTPS webhooks into clean, source-branded iPhone notifications. Each service gets its own name, avatar, destination URL, and secret endpoint. Intended senders: CI, agents, scripts, monitors — anything that can POST JSON. Hosted product: https://hark.ryan.ceo Public monorepo: https://github.com/R44VC0RP/hark (TypeScript; Node 22+; compose.yaml / Dockerfile present in tree)
Is Agentic is a free, public Vercel product that scores how ready a public HTTP or HTTPS website is for AI agents. Submit a URL on the home page. Ora runs and scores every technical check. Is Agentic groups those checks by maturity and applicability, then publishes a score, evidence, recommended fixes, and a shareable report.
Magika (google/magika) is Google’s AI-powered file content-type detector: it classifies files from byte content (not just extensions / magic bytes), locally, in milliseconds on CPU. First-party README: ~100M-sample training, 200+ content types, ~99% avg precision/recall on their test set; used at scale for routing Gmail/Drive/Safe Browsing samples to scanners; also integrated with VirusTotal and abuse.ch (README claims).
README: "Mailcheap is a self-hosted newsletter app on Amazon SES. Write an issue, send it to your list, publish it on the web. About $0.10 per 1,000 emails." One small Next.js app with an editor, list, signup forms, unsubscribe, bounce and complaint handling, stats and a public archive.
OpenDisk is a native macOS disk-space analyzer. README positions it as a fast, free, open-source DaisyDisk alternative: whole drive as an interactive sunburst (hover size, click to zoom), sortable Finder-style list, live streaming scan, incremental rescans, APFS volume groups / firmlinks / purgeable awareness, Collector (drag then delete), Sparkle auto-update.
OpenRouter is a unified large language model inference router and model marketplace: one OpenAI-compatible HTTP application programming interface across many underlying inference providers, with automatic fallbacks and cost-aware routing. The marketplace object is models and inference providers (tokens in → tokens out), not installable agent tools ranked by dual human-and-agent reputation.
A mature AGPL-3.0 open-source agent-control surface: a local Paseo daemon runs and manages coding-agent CLIs on a user-controlled machine, while desktop, mobile, web, and CLI clients monitor/control those agents. It supports Claude Code, Codex, Copilot, OpenCode, Pi, and 30+ other providers/integrations according to its site.
Prime Agent is an MIT-licensed open-source coding and research agent from Prime Intellect, built around a Recursive Language Model (RLM) runtime and a persistent IPython control environment. The model works through programmatic Python: file/shell operations, data transforms, skills, context management, and recursive subagents are all callable from the durable kernel rather than exposed as a large fixed tool palette.
Refero is a design-research platform of real web and iOS product interfaces, user flows, UI patterns, and visual styles. Its Refero MCP exposes structured research to agents so they can study relevant screens and flows before implementation.
Search is a macOS browser by Office Commun that deliberately ships without its own engine: it uses the system WebKit (same family as Safari), which is why the app footprint stays in the low-megabyte range.
Docs intro: shadcn/ui is a set of beautifully-designed, accessible components and a code distribution platform. Works with your favorite frameworks and AI models. Repo README: Use this to build your own component library. Docs principles: Open Code (you get the component source), Composition (shared composable interface), Distribution (flat-file schema + CLI), Beautiful Defaults, AI-Ready (Open code for LLMs to read, understand, and improve).
Soar is a consumer flight product (find/book cheap flights) shipping a Flight Booking MCP so agents can search, book, pay, and manage trips (plus public stay search). Positioning vs many flight MCPs: book + pay, not search-only. Launch (@SoarAI, 2026-07-26): > Today we’re releasing the first MCP to find and book flights, always at market rates.
Hosted observability for AI/LLM apps. Product line: every model call, tool step, and retrieval in one OpenTelemetry trace, with tokens, cost, latency, and errors. Ingest via TypeScript or Python SDKs, or any OTLP exporter. Cost is computed server-side against 4700+ models.
tldraw offline is tldraw’s local, file-based desktop infinite canvas (no accounts / no server required for core use). Documents save as portable .tldraw files (canvas + media + optional embedded document scripts). The 2026-08-03 @tldraw post demos the agent-native angle: you can make your own tools on the canvas — e.g.
turbopuffer is a managed, commercial search engine from turbopuffer Inc. (Canadian, founded 2023). It is object-storage-first: Amazon S3 / Google Cloud Storage / Azure Blob is the source of truth; NVMe and memory are caches. It does approximate vector search (SPFresh), BM25 full-text, filters, and hybrid fusion.
Clapper is an MIT-licensed open-source React → MP4 video tool from ArchAstro, built by Calvin Grunewald (@CalvinGrunewald). Site byline: Open source, MIT, by Calvin at ArchAstro. Tagline: Write videos in React. Review them like a studio would.
OpenAI Codex Security plugin (codex-security v0.1.11 in plugin.json at capture): reusable Codex workflows for security scans, analysis, validation, investigation across codebases and diffs.
Apple container is a tool for creating and running Linux containers as lightweight virtual machines on a Mac. It is written in Swift, optimized for Apple Silicon, and consumes/produces OCI-compatible container images. The project uses Apple's lower-level apple/containerization Swift package for container, image, and process management.
emulators.dev is a hosted catalog of per-service emulator hosts. Landing subtitle: stateful integration emulators for real developer APIs; each emulator has its own host; open one or fetch its manifest. Bar links Source Code / Report Issue to vercel-labs/emulate and Learn More to https://emulate.dev.
Generative Loaders is an MIT-licensed React package and live gallery for loading states specific to generative interfaces: streaming-text reveals, compact inline activity indicators, and image-generation placeholders. Install with npm install generative-loaders; it supports React 18+ and uses Framer Motion.
Treg is an agent-oriented tool registry and credential-injection proxy. Its hosted catalog claims roughly 2,600 endpoints across about 40 providers, searchable by task with request/response shape and per-call prices. A team can also register its own API credentials, OAuth connections, CLIs, and SKILL.md bundles; callers use one Treg token while upstream credentials remain server-side.
VictoriaMetrics develops separate storage/query components for the three primary OpenTelemetry signals: - VictoriaMetrics for time-series metrics; - VictoriaLogs for logs; and - VictoriaTraces for traces. Its current documentation says each component accepts OpenTelemetry data natively and can sit behind instrumented applications or an OpenTelemetry Collector.
Larasend is an open-source, self-hosted transactional email platform for Laravel teams. It provides a clean HTTP API (POST /api/emails), dashboard, activity log, scoped API keys, delivery events, suppressions, rendered previews, DKIM/DNS guidance, and a Laravel mail transport while sending through your own Amazon SES account.
A design-reference workflow built around Mobbin plus MCP access, useful for feeding high-quality UI examples into implementation context while building.
Cloudflare Monetization Gateway (announced 2026-07-01) is an engine for Cloudflare customers to charge for any asset protected by Cloudflare: web pages, datasets, APIs, or MCP tools. Single control plane for payment policies and access controls; payment verification and enforcement at the edge so origins are not hammered by payment-volume traffic.
Railcode is Yakko Majuri’s secure / governed cloud for internal software — apps and agents (“internal Vercel” framing on the launch post). Build with your own tools; deploy into a company-governed sandbox with auth, permissions, data, and connectors centralized.
README tagline: "One MCP for reverse engineering across binaries, applications, and runtime behavior." The agent calls REA through MCP to inspect a local target and trace relevant code; REA returns findings with their evidence, and the agent uses them to explain the behavior or write and test an implementation.
Rikyū (named for the tea master) is an AI design SaaS: describe what you need in a sentence and it generates a coherent brand/asset set — logo, web, social, print — carried across one visual identity. Positioning: "skip the design work" — no design skills, no asset hunting, one design language across everything that follows.
@shinzolabs/gmail-mcp is Shinzo Labs’ local MCP server for Gmail API access (OAuth2, file-based credentials). It ships anonymous product telemetry on by default via env TELEMETRY_ENABLED (default true). Opt out by setting TELEMETRY_ENABLED=false in the MCP client env block, then restart the client/server.
Kubernetes-native platform for isolated, stateful singleton workloads used as AI agent runtimes and development environments. Sandbox, SandboxTemplate, SandboxClaim, and SandboxWarmPool CRDs provide a declarative API for long-running containers with stable identity, optional pers
App Store Connect CLI (asc) — fast, JSON-first Go CLI for the App Store Connect API. Automates TestFlight, builds/uploads, submissions, signing, analytics, screenshots, subscriptions, Apple Ads, and related release workflows for iOS/macOS/tvOS/visionOS.
Evil Buttons (site title: Evil Buttons — Animated shadcn/ui components with Motion) is a shadcn/ui registry of playful, highly animated button components with an “evil touch” aesthetic. Site claims 31 interactive components, live previews, copy-paste docs, and one-command CLI installs.
MIT-licensed Node CLI + agent skill by Peter Yang. It opens a target HTML/Markdown file or localhost route in a local browser review UI. The reviewer can directly change text/basic formatting, resize images, delete elements, and attach comments to selected text or blocks.
Portable agent skill that edits drafts to remove 20+ AI-slop patterns (min effective edit + self-check via eval.md) or detects patterns with quoted lines — without claiming AI-authorship detectors.
Landing title Rare UI — Rare Animated React Components. H1: Tasteful Components, Made to Stand Out. Body: a free, open-source collection of rare animated React components. Install any item with the shadcn CLI. README: a shadcn registry built with Next.js, Tailwind CSS, and TypeScript.
Slack feature that searches conversations, files, connected third-party apps, and custom or self-hosted sources from one Slack search bar. It uses real-time federated retrieval, permission-aware results, and natural-language answers with conversational summaries. Slackbot is desc
Alex is a Rust-based local LLM proxy and trace control plane created by Madhava Jay. It sits between supported AI harnesses and model providers/subscriptions, translates several model API dialects, manages provider credentials and account selection, and stores captured traces in a local SQLite-backed system.
Amicro is an open-source React catalog for micro-interactions, entrances, hovers, cursor effects, loaders, card spreads, 3D carousels, charts, and other motion-heavy UI patterns. The repository describes the components as powered by Motion and supports live previews, source copying, raw shadcn registry items, and an npm package.
Local-first AI agent whose control loop and all state run on the machine. Drives browser, files, approved shell, documents, memory across sessions, scheduled tasks, and MCP tools. Default path is llama.cpp (their TurboQuant fork) so small quantized models stay useful for long multi-step work on consumer hardware.
BetterWright is materially faster once warm, slightly more compact, and has a strong agent-oriented design: persistent sessions, code snippets, diff snapshots, credential redaction/vaulting, network policy, evidence artifacts, and an optional contained agent loop.
Site and README title beUI v2. Tagline: “Animated components for React and Next.js. Copy the source, own the code.” README: “a small component library for product interfaces.” Each item has a live preview, usage example, source, and a shadcn install command.
First-party landing: one command turns a laptop server into a public, encrypted URL on Cloudflare’s edge — no account, no DNS, no open inbound ports. cloudflared opens an outbound-only connection; traffic to a random .trycloudflare.com hostname rides Cloudflare’s network back to localhost.
Exact author text: > We are open sourcing cn, a new engine for Tailwind class merging and conflict resolution. It is a drop-in replacement for tailwind-merge and clsx. > > It works with all frameworks, runs everywhere and is 30x faster. > > Brought to you by @aidenybai and me.
Company Brain (Apache 2.0, TypeScript) is supermemory’s former paid Slack product, open-sourced and rebuilt to run on your own Cloudflare Workers account (D1, KV, Durable Objects, Workers AI). README pitch: A teammate in your Slack that truly knows and understands your company, and can do anything.
GitHub description: An open source agent microharness featuring persistent agency and recursive LLMs. Of bash, by bash, for bash; it's shells all the way down. Homepage field: headlong.ai. README: complete harness with a core of less than 10K lines of Bash; defining feature is persistent agency — the agent keeps thinking between external interactions in a self-guided loop.
Hex calls itself "the AI Analytics Platform where trust meets insight": one workspace where data practitioners run deep analysis (SQL, Python, no-code in agentic notebooks), business users ask questions in plain language, and both work on the same data.
Laya = Apache-2.0, non-autoregressive System One decision model family: given a state (text / email / ticket / JSON) and typed questions (choice / score / noul), returns typed answers with probabilities in one forward pass. Trained with RLCD (reinforcement learning against strictly proper scoring rules).
LittleJS is a tiny, fast, dependency-light HTML5/JS 2D game engine (WebGL2 + Canvas2D hybrid, particles, sound, input, etc.). Author Frank Force (@KilledByAPixel). Repo includes demos, docs, FAQ, JS13k-oriented branch, CLAUDE.md for agent-assisted work.
Mono Charts is a set of 30 minimalist chart previews with monochrome palettes, rounded geometry, compact typography, hover states, and selected entrance/data animations. The live tab links to a dedicated TypeScript repository containing the catalog page, chart metadata, and component source.
Open Notebook is an open-source implementation of Google NotebookLM-style workflows with more model choice and deployment flexibility. The project positions itself as a private, multi-model, local/self-hostable notebook for ingesting sources, searching them, chatting with grounded context, and generating podcasts from research material.
OpenShaders is a pre-launch, MIT-licensed platform being built as a home for shaders aimed at real websites — publish, discover, upvote, follow, and install — not demos-only experiments. Site title: OpenShaders: Shaders belong on the web. Hero: Shaders belong on the web.
The computer underneath the agent. Instant full cloud desktops (GUI, browser, filesystem, terminal, apps) that AI agents can see and control via API — boot claimed under ~500ms. Always-on until stopped. Linux + Windows GA; Mac closed beta. People use it three common ways (docs): 1.
Pencil / pen.dev (site banner: “pencil.dev is now pen.dev”) is an AI-native infinite design canvas positioned as “Dream on canvas. Land in code.” Backed by a16z Speedrun (site badge). Integrates with Cursor, VS Code, Claude Code, OpenAI Codex; emphasizes bi-directional MCP vector canvas, open file format, Figma bring-over, brand kits, and design-as-code.
pgbot is a read-only, findings-first PostgreSQL diagnostic written in Go. Official README: one static binary connects, reads Postgres's own statistics views, and prints a graded health report plus what changed since last time. No collector, no time-series store, no write privilege anywhere in the path.
QuiverAI ships foundational models for generating, editing, and animating editable vector/SVG assets (logos, illustrations, diagrams, line work; typography & animations marked coming soon on marketing site). Arrow 2 is the Sep 2026 model line called out as higher quality and faster, in App and API.
Smaug (MIT, JavaScript) archives Twitter/X bookmarks and optionally likes to markdown. README pitch: Like a dragon hoarding treasure, Smaug collects the valuable things you bookmark and like. Pipeline (README): 1. Fetch bookmarks/likes via bird CLI (Peter Steinberger) using browser session cookies 2.
Personal playground of CSS animation studies by yui540. Repo description: personal CSS animation playground (Japanese original on the GitHub description). README title: CSS animation is fun. Body: they publish some of the CSS animations they played with.
Agency is a local card-feed app from Browser Use: a coding-agent coordinator prepares suggestion cards; the human approves with one click (Send / Apply / Merge). Thesis (X + README): agents find useful work; you decide — “Never prompt again.
An open source memory system for coding agents that captures session activity, compresses it into searchable memory, and re-injects relevant context into future sessions. The repo positions it as agent-agnostic infrastructure that can work across Claude Code, OpenClaw, Codex CLI, Cursor, Gemini CLI, and other MCP or HTTP-capable agents.
CUA-S1 = trycua’s open-source line of small “System One” specialists for computer use (not general chat). First public member CUA-S1-FORMS: a ~706k-param one-pass option scorer that, given a UI element context + typed options (fill … / check / click / skip), returns a probability per option in one forward pass.
Design Spells is a browsable visual reference library for small interface details—especially motion, interaction states, easter eggs, transitions, buttons, dynamic-island behavior, and 3D/“fun” product moments. It organizes examples by both tags and source product/app, with individual “spell” entries that pair a recorded interaction with lightweight taxonomy.
Needle 3 is Cactus Compute’s on-device “automation foundation model”: a laddered / sliceable tiny model (post: 8–29 MB CQ2-bit binary; 25–121M params; depths 2–20 layers from one weight set) built on their Simple Attention Networks, aimed at tool calls, structured extraction, and local text embeddings — not open-ended chat.
Open-source enterprise search and AI assistant that connects to internal knowledge sources, indexes them with source permissions, and supplies humans and agents with answers and context. It is described as self-hosted in the customer's cloud, VPC, Kubernetes, or air-gapped enviro
Aval is a web-only format + runtime for short prerendered animation: continuous partial loops, user-defined states, authored triggers, bounded transitions, packed transparency, and host-owned fallback markup. Central idea (README): encode independently decodable motion units and a small deterministic state graph in one .avl asset.
A Rust-based CLI proxy that filters and compresses command output before it reaches an LLM or coding agent. The repo positions RTK as a low-overhead way to reduce token consumption on common development commands like git status, git diff, test runners, grep/search, logs, and cloud/devops tooling.
Private voice journal for iPhone with one rule: nothing leaves your phone (optional encrypted iCloud backup only if user enables it). Speak → on-device transcript → on-device “reflection” (topic / how it sat / one observation). Local model is an observer — no advice, comfort, or reply.
Site title Transitions.dev: UI transitions for AI agents. README: “an interactive collection of reusable CSS transitions. Each card on the index page demonstrates a different interaction pattern and ships a copy-ready, portable CSS snippet.” Live index groups cards as All / Essential / AI Agents / Effects / Texts / Pro.
Vellum Assistant is a personal AI assistant product (OSS + managed cloud) in the same broad class as Hermes / OpenClaw / NanoClaw: multi-channel agent, skills, memory, computer use, OAuth-connected tools, managed or self-hosted runtime. Entry capture was the OAuth Integrations key-concept doc: browser OAuth2 instead of pasting API keys; each integration is a bundled skill with tools (e.g.
