Agenta delivers a comprehensive open-source LLMOps workspace that covers the full lifecycle of AI application development — from prompt engineering through production monitoring. The platform supports 15+ model providers including OpenAI, Anthropic, Google Gemini, Mistral, Groq, Together AI, Azure, AWS Bedrock, and self-hosted models via Ollama, enabling teams to switch between providers without code changes. The prompt playground allows side-by-side comparison of different configurations, while the evaluation system offers LLM-as-a-Judge assessment, 20+ pre-built evaluators covering semantic similarity, regex matching, and factual accuracy, plus custom Python evaluators for domain-specific requirements. Teams run evaluations through both the web UI for subject matter experts and the Evaluation SDK for programmatic CI/CD integration. The observability layer captures full trace visibility across complex agentic workflows, flagging quality issues like hallucinations and off-topic responses in real time. Human annotation workflows let domain experts review and annotate LLM outputs, feeding corrections back into the evaluation loop. The architecture supports Chain of Prompts, RAG pipelines, and multi-step agent workflows, integrating with frameworks like LangChain and LlamaIndex. Self-hosting deploys via Docker Compose with Traefik for routing, requiring only a clone, environment configuration, and a single docker compose command. On RepoCloud, deploy Agenta on a dedicated VPS with root SSH access, persistent storage for evaluation datasets and traces, and complete control over model provider credentials, all under the MIT license with no usage restrictions.

One binary is all that stands between you and every Kubernetes cluster you own: Kubewall opens a realtime web dashboard over all of them, with no agents to install and nothing phoning home. Switch between clusters from the sidebar and browse workloads, networking, storage, configuration, and access control in tables that update live as pods start, fail, and reschedule. Open any resource to inspect its overview, edit its manifest in the built-in YAML editor with validation, watch its events, or stream logs aggregated across pods and containers with search and tail controls. A command palette jumps straight to any resource type, while faceted filters and global search cut through thousands of objects by name, namespace, label, or image. Scale deployments, restart pods, trigger rollout restarts, and apply manifests with a click, or forward a cluster service to your local machine without memorising kubectl incantations. An integrated AI assistant, connected to OpenAI, Claude, Gemini, DeepSeek, OpenRouter, Ollama, or a local model, helps troubleshoot failures and review configurations right beside the resource it concerns. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Private machines hide their services behind routers and firewalls; sish gives those services a public address, reached with nothing more exotic than the SSH client already on your laptop. Point a standard remote forward at the server and a local web app appears at its own subdomain over HTTPS, ready to share with testers, clients, or a webhook provider. Raw TCP forwards work the same way, publishing a database, a game server, or an SSH port on a fixed external port you choose. Private aliases keep sensitive services off the public internet entirely, reachable only through an authenticated SSH jump from another machine you control. SNI routing passes encrypted TLS connections to the right backend by hostname, so several services share one public endpoint without the tunnel terminating their encryption. Aliases can spread traffic across multiple backends with built-in load balancing modes, handy when several replicas sit behind one name. The service console lists live tunnels and inspects forwarded requests as they pass, turning tunnel debugging into reading a log instead of guessing. Access rests on SSH keys or passwords with hot-reloaded key directories, and binding policies cap which subdomains and ports each user may claim. Because clients need no agent software, locked-down laptops, CI runners, and embedded devices can open tunnels with the tools already installed. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Define a DocType in Frappe and the framework automatically generates database tables, Python ORM classes, REST API endpoints, form views, list views, and role-based permission rules from that single schema definition. The meta-driven architecture stores schemas as database records, letting administrators add fields, reorder layouts, and modify validation rules through the browser-based Form Builder without redeployment. Server-side Python controllers handle business logic through hook-based lifecycle events while client-side JavaScript manages interactive form behavior, with WebSocket connections providing real-time updates across sessions. The built-in workflow engine defines document state machines with role-gated transitions, email notifications, and conditional action visibility. Pre-configured Desk views include list, form, report, tree, kanban, calendar, and dashboard layouts with drag-and-drop workspace customization. The report builder generates grid reports with configurable columns, filters, grouping, and chart visualizations from DocType data or custom SQL and Python scripts. Background job processing uses Redis queues for email delivery, data imports, and periodic tasks. Virtual DocTypes connect external databases to the framework's UI without data migration. Docker deployment via frappe_docker provides production-ready Compose configurations with MariaDB, Redis, Nginx, and worker containers. The foundation powering ERPNext, HRMS, CRM, and Helpdesk in production. Over 10,500 GitHub stars. MIT licensed.

GitHub Release Monitor turns release day into a notification instead of a discovery, watching GitHub, GitLab, Codeberg, and self-hosted Forgejo repositories in the background and reporting each new release to the channels you already read. Your watchlist is the whole interface. Add the projects your stack depends on, and scheduled checks compare what is published against what you have seen. Email carries the full notes over SMTP. Apprise carries them further, to Discord, Telegram, Slack, and dozens of other services, either one message per release or bundled into a digest for each check. Filtering happens at two levels. Global rules decide which release types count, and any repository can override them with its own channels, its own include and exclude regular expressions on tags, and its own choice of alpha, beta, or release candidate lines. A pattern the current engine no longer supports is flagged in settings rather than silently passing nothing. The list itself is portable JSON, ready to export from one install and import into another. When the alerts stop, the diagnostics page separates suspicion from evidence by testing forge connectivity, SMTP, and Apprise in one place. Accounts protect the app, dark mode and nineteen languages shape it, and the layout works on a phone. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. AGPL-3.0 licensed.

With over 77,000 GitHub stars, Crawl4AI is the most-starred open-source web crawler on GitHub and the go-to tool for converting the web into AI-ready data. Built on Python and Playwright, it transforms any website into clean Markdown with headings, tables, code blocks, and citation hints optimized for LLM ingestion, or extracts structured JSON via CSS selectors, XPath expressions, or direct LLM-based schema extraction through OpenAI, Anthropic, and Ollama providers. The self-hosted Docker server exposes a REST API on port 11235 with endpoints for crawling, streaming results, screenshots, PDF generation, JavaScript execution, and LLM-powered extraction. Version 0.9.x introduced secure-by-default operation with mandatory JWT authentication, strict request validation, declarative hooks replacing inline code, and bounded job queues. Adaptive crawling uses information foraging algorithms to determine when sufficient data has been gathered, while deep crawl mode traverses link graphs intelligently. The async browser pool manages concurrent sessions with stealth plugins, proxy rotation, custom headers, and session persistence for authenticated scraping. A built-in MCP server enables direct integration with Claude, ChatGPT, and Cursor for AI-driven web research workflows. Content filtering applies BM25 and TF-IDF relevance scoring to extract only pertinent sections from noisy pages. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache 2.0 licensed.

GitHub forgets your traffic after fourteen days; ghstats remembers it for years. This self-hosted dashboard polls the GitHub API every hour and keeps views, clones, stars, forks, issues, and pull requests for every repository in its own database, long after the built-in traffic tab has rolled over. Open a repository to see its charts for stars, clones, and views next to top referrers and popular paths, filter the list by owner, name, forks, or archived status, and include private repositories when your token allows it. Public stats pages and portfolios reuse the same numbers through a JSON API, embeddable SVG badges for views, clones, and star velocity, and cards ranking your top repositories or showing star history across projects. Release download counts accumulate as daily snapshots too, so a launch spike stays visible months later. The whole service is one small Docker image on port 8080 with a single file for storage, easy to back up and easy to move. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

When your services need to talk without waiting on each other, Apache RocketMQ carries the conversation: a distributed messaging platform where producers publish to topics and consumers process at their own pace, with delivery guarantees you can build money movement on. Send ordinary messages for fire-and-forget events, transactional messages that commit or roll back with your local database transaction, ordered and FIFO messages for sequences that must not shuffle, and delayed messages that arrive seconds or days later on schedule. Consumers pull with long polling, filter by tag or SQL expression at the broker, and retry failures through built-in retry queues and dead letter queues instead of losing work. Clusters scale by adding brokers, while Controller mode elects broker replicas automatically so a failed node stops being a page. The Proxy component speaks gRPC with modern multi-language clients and adds POP consumption, where messages become invisible while processed and reappear if the consumer never acknowledges. Tiered storage offloads cold messages to cheaper object storage while hot data stays on local disk, and RocketMQ Streams, Connect, and an MQTT bridge extend the same cluster into stream processing, data integration, and device messaging. Lite Mode trims subscription overhead for AI workloads, and the separate RocketMQ Dashboard gives operators a web console for topics, consumers, and message tracing. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Obscura is the headless browser you point Puppeteer or Playwright at when Chromium feels like renting a truck to fetch a letter. Start the server and your existing automation connects over the usual DevTools protocol, opens pages, clicks, types, waits for selectors, and reads the fully rendered result, JavaScript included. Scrape one page or fan out across many in parallel, evaluating an expression on each and getting structured output back. Capture viewport or full-page screenshots, export PDFs, intercept requests to block or rewrite them, and manage cookies and headers per session. An MCP mode lets AI agents drive the browser as a tool instead of shelling out to scripts. Stealth mode randomizes fingerprints per session, presents realistic browser identity signals, and blocks thousands of known tracker domains before they load. Proxies, persistent storage for cookies and local storage, and robots handling round out the controls for longer crawls. The whole thing starts instantly and sips memory compared with a conventional headless Chrome, so a modest server can keep many more pages in flight. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Universal database administration across modern enterprise infrastructure is simplified through dbx, a lightweight web client and SQL studio supporting over ninety database engines. Database administrators connect to MySQL, PostgreSQL, SQLite, Redis, MongoDB, ClickHouse, DuckDB, Oracle, and Microsoft SQL Server through authenticated network connections and encrypted SSH tunnels. The interactive query editor provides real-time multi-schema autocomplete, syntax linting, execution plan analysis, and natural language SQL query synthesis powered by configurable large language model backends. Teams visualize complex relational database architectures using interactive entity-relationship diagrams that map table foreign key constraints, primary indexes, and table partition layouts automatically. Operators inspect live Redis memory usage breakdowns, monitor real-time server command throughput, and manage key expirations without opening raw terminal sessions. The editable data grid supports multi-row modifications, foreign key reference lookups, CSV and JSON exports, and binary blob data inspection directly inside the browser viewport. Software developers expose connected databases directly to external AI coding assistants through native Model Context Protocol server endpoints. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Alma Relay is the relay you run yourself so your phone can reach the Alma on your desktop from anywhere, without port forwarding, a VPN, or a third party sitting in between. Your desktop dials out to the relay and keeps that connection open, and when you open the app on your phone, its requests travel through the relay to the desktop and the answers stream back the same way. Pairing happens with a QR code that carries the relay address, so switching to your own relay takes one fresh scan instead of manual setup on every device. New pairings encrypt the session end to end, so the relay routes encrypted traffic it cannot read, and the routing values it handles are one-way hashes that are never written to disk or logs. When your phone is locked, the relay can pass along the "reply finished" and "Alma has a question" notifications through Apple's or Google's push services, so you hear about finished work without staring at the screen. A live view of the desktop's browser can ride along too, with short-lived credentials issued for the media connection. Sharing the relay with a team works through per-person tokens you can revoke individually, and regenerating your connect code cuts off every paired phone at once. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Aegra runs your LangGraph agents on your own infrastructure with the same SDK and APIs you would use on the managed platform, so moving in does not mean rewriting your agents. You deploy an agent once and clients connect through Agent Chat UI, LangGraph Studio, or CopilotKit, while threads, runs, and checkpoints persist between sessions instead of vanishing when a process restarts. Streaming answers arrive token by token over server-sent events, and if a connection drops, the stream reconnects and replays missed events rather than losing the run. Human-in-the-loop gates pause an agent at the step you choose, wait for a person to approve or edit, and then resume from exactly that state. Scheduled cron jobs trigger runs on standard or seconds-level expressions with timezone support, which turns agents into recurring workers for reports, checks, and follow-ups. When traffic grows, Redis-backed workers process runs with lease-based crash recovery, so a dead worker's job is picked up instead of disappearing. Authentication handlers plug in JWT, OAuth, or Firebase, custom endpoints sit beside the agent API, and traces fan out to the observability backend you already run. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Searcharvester gives your AI agents a private search and research backend that runs entirely on your server, with a web UI where you watch every research job unfold. The search endpoint speaks the Tavily schema, so existing agent code points at your host and queries more than a hundred SearXNG engines without API keys or quotas. Send a URL to the extract endpoint and it returns clean markdown with navigation, footers, and ads stripped, sized in presets from five thousand characters to fully paginated long documents. Three extractors compete on each page and a quality gate keeps the best result, so hard pages still yield readable text. For bigger questions, the research endpoint runs a team of agents: a lead plans sub-questions, researchers gather in parallel, a critic tries to disprove the findings, and a fact-checker re-verifies numbers and dates before the final report arrives with clickable citations. You assign a different model and reasoning level to each role, follow agent branches, the reading graph, sources, and media live, then read the finished report in the same UI. Search settings let you toggle engines per category, test proxies, and apply changes with a targeted restart. A loop guard enforces budgets for calls, tokens, searches, and page reads across the whole job. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. AGPL-3.0 licensed.

Your AI coding agent has your API keys in its environment and unrestricted network access, which means one prompt injection away from sending those secrets anywhere. Pipelock closes that gap by sitting as a proxy between your agents and every outbound connection, scanning the actual content of HTTP, WebSocket, MCP, and Agent-to-Agent traffic before it leaves your server. An 11-layer scanner pipeline checks every request against 62 credential patterns covering AWS, GCP, Azure, GitHub, OpenAI, Anthropic, SSH keys, and database URLs, then inspects every response for prompt injection using 29 detection patterns with six-pass normalization that catches base64-encoded, leetspeak, and whitespace-obfuscated payloads. The MCP proxy wraps any Model Context Protocol server (stdio, HTTP, or WebSocket) with bidirectional scanning that detects tool description poisoning and mid-session rug-pull changes via SHA-256 fingerprinting. Every scanning decision produces a cryptographically signed action receipt that third parties can verify offline without trusting the agent or the vendor. The Operator Console provides a web dashboard for reviewing evidence scorecards, receipt timelines, agent sessions, enforcement decisions, and fleet posture at a glance. Cross-request taint tracking catches slow-drip exfiltration attempts that spread a secret across multiple calls. Canary tokens plant synthetic secrets that trip alerts the moment an agent tries to exfiltrate them. Pre-built Prometheus metrics and a Grafana dashboard provide real-time visibility into traffic volumes and block rates. Deploy on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache 2.0 licensed.

When a product needs sign-in, teams, permissions, and user analytics, Hexclave (formerly Stack Auth) delivers all of it from one admin dashboard, so nobody on your team hand-rolls another password reset flow. Sign-in methods are switches, each flipped on or off without a redeploy: email and password with secure reset, magic links, passkeys, TOTP two-factor, any of 12 OAuth providers including Google, GitHub, Microsoft, Apple, and Discord, or your own OpenID Connect provider. Sign-up rules allow, reject, or hold accounts by email domain, country, auth method, or fraud risk score. The user directory searches, filters, edits, and exports every profile, custom metadata included. Teams bring workspaces, email invites, and nested roles checked the same way on server and client. API keys belong to users or teams, auto-revoke when leaked, and reveal their secret only once. Themed templates send transactional and marketing email with open and click tracking. Analytics show live active users and session replays and store the SQL queries you write, and every user event can fire a signed webhook with automatic retries. Standard JWTs verify locally against a JWKS endpoint, and SDKs cover Next.js, React, and plain JavaScript. Self-host the dashboard and API on your own server and export your users whenever you like, with no lock-in. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT and AGPLv3 licensed.

AGenUI Studio turns a written description into a working interface card: an AI agent generates the UI, binds it to your APIs and data, and packages it so it runs anywhere, no Studio required. Bring your own model by entering any OpenAI-compatible or Anthropic Messages endpoint and key in the console settings, and generation proceeds contract-first, steered by design rules you publish as versioned Markdown revisions. You preview each card live as it is generated, then refine it over multiple turns while every component, binding, and action you did not target stays protected from accidental edits. The console also manages the supporting cast: a local knowledge base, API definitions, and reusable operators, all starting from bundled demo data you can swap for real services whenever you are ready. When a card is finished, download it as a self-contained Runtime Package carrying its data sources, bindings, actions, and operators, and execute it with the bundled Card Runtime inside your own application, or register a delivery target and publish straight from the workbench over a callback or a message queue. Applications subscribe to native run events over HTTP and SSE, and production deployments can replace the local defaults piece by piece, MySQL for storage, Redis for hot state, MCP services for knowledge and data, while the packages keep the same contracts. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

With over 2,700 GitHub stars, Auto Company is the first open-source framework that runs a fully autonomous AI company 24/7 — 14 specialized agents modeled after Jeff Bezos (CEO strategy), Werner Vogels (CTO architecture), Charlie Munger (critical analysis), DHH (full-stack engineering), Kelsey Hightower (DevOps), Seth Godin (marketing), and eight more domain experts collaborate through dynamic squad formation to ideate products, write code, deploy infrastructure, and execute marketing campaigns without human intervention. The five-layer architecture separates execution, orchestration, cognition, workflow routing, and observability, while the consensus memory pattern uses a single markdown file as a relay baton between cycles — no vector databases, no Redis, no embeddings required. A bash loop invokes Claude Code or OpenAI Codex CLI every 30 seconds, each cycle selecting 2-5 agents from the 14-person pool based on task context. Over 30 reusable skills handle specialized tasks from frontend design to competitive analysis and deployment automation. Circuit breakers trigger cooldown after consecutive errors, rate-limit detection auto-sleeps on API throttling, and sandbox rollback protects against destructive changes. The Python-powered web dashboard displays real-time cycle status, cost tracking per cycle averaging under $2, and agent activity visualization, with CLI control via make start, stop, monitor, pause, and resume. Supports macOS via launchd, Windows via WSL with systemd, and native Linux deployment. The npx create-auto-co command scaffolds a new AI company in seconds. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Admin panels, database GUIs, dashboards, approval flows, customer support consoles - Appsmith builds the internal tools your team keeps postponing, on an open-source low-code platform. The UI assembles from 45+ drag-and-drop widgets - tables with server-side pagination and inline editing, charts, forms, lists, buttons - which bind to data through {{ }} JavaScript expressions anywhere in the editor. Datasources cover PostgreSQL, MySQL, MongoDB, MS SQL, Redis, Snowflake, and more, plus any REST or GraphQL API, with SaaS integrations and AI query support for prompt-based steps inside apps. When the widget library falls short, custom widgets are plain JavaScript, HTML, and CSS, and external JS libraries can be imported, which keeps the platform extensible where pure no-code tools hit walls. Git-based version control enables branch-based collaboration, review, and rollback of app definitions. Queries and JS objects hold the business logic layer between datasources and UI. Self-hosted via Docker or Kubernetes, with role-based access control for published apps.