QwenPaw
Designed as a unified personal AI workstation, QwenPaw operates as an autonomous digital copilot that coordinates scheduled automations, interactive coding sessions, and multi-channel team communication across everyday messaging apps. Users can dispatch long-running research tasks, process office documents including PDF, Excel, and Word files, and execute browser-based data collection without manual intervention. The built-in web console and terminal interface provide full visibility into agent reasoning, allowing operators to inspect intermediate thinking steps, review proposed file edits, and approve sensitive tool calls. Through direct connectors for Discord, Telegram, DingTalk, Lark, and WeChat, teams can trigger specialized skills or query shared workspaces directly from their existing group channels. A self-evolving personal memory system continuously indexes chat interactions and local resources into editable Markdown files, ensuring knowledge persists across restarts and task delegations. Built-in guardrails including a sandboxed execution runtime, tool access policies, and automated skill scanners protect underlying host files from unauthorized modifications during autonomous scripting runs. 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.
GPT Image Playground
One prompt box, four model families, and a gallery that never leaves your browser: GPT Image Playground turns any compatible image API into a tidy studio for text-to-image and reference-based editing. Attach reference photos by paste, drag, or link, then paint a mask or sketch notes directly on the picture to show the model what to change. Agent mode keeps a multi-turn conversation going, calls the image tool as needed, generates related pictures in parallel, and branches whenever you edit and resend a message. Every task is saved with its size, quality, elapsed time, and the prompt as the model actually rewrote it, so you can compare what you asked for with what happened. Favorites sort into several collections that download as ZIP archives, and failed jobs retry in one click. Saved API configurations switch from a dropdown, covering OpenAI-compatible endpoints, Google, xAI, ByteDance, fal.ai, and custom HTTP providers. Output formats include PNG, WebP, animated GIF, and SVG, with transparent backgrounds on request. Operators can preload configurations through environment variables and route calls through a built-in proxy. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
E2A
E2A is an open-source email API that lets any application send transactional email and gives AI agents real two-way inboxes of their own, so an assistant can hold a conversation over email without borrowing a person's mailbox. Register an agent and it gets a working address on a shared domain in seconds, or verify your own domain for branded addresses, then connect through the hosted MCP server, a coding-agent plugin with browser OAuth, signed webhooks, a WebSocket stream, REST polling, or the TypeScript and Python SDKs. A local agent behind a firewall needs no public URL at all: it listens over WebSocket and replies in-thread with conversation identifiers keeping multi-turn exchanges coherent across humans and other agents. Every inbound message arrives with structured SPF, DKIM, and DMARC evidence, with the DMARC alignment verdict included, so your agent can judge how far to trust a sender instead of trusting the From header. Flip on the human-in-the-loop gate per agent and outbound messages pause for a person to approve, edit, or reject from the dashboard, the SDK, MCP tools, or magic links in their own inbox, with expiry rules and recipient allowlists narrowing which sends get held. Inbound content screening flags prompt-injection payloads before the agent ever sees them. Support bots triage intake, scheduling agents coordinate meetings, and commerce agents answer order questions from one persistent address. 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.
Novu
Powering notification delivery for thousands of organizations with over 39,000 GitHub stars, Novu is the open-source communication infrastructure that eliminates the need to build separate integrations for every messaging channel. The TypeScript-based platform provides a single API endpoint that routes notifications across In-App Inbox, Email via SendGrid, Mailgun, AWS SES, and Postmark, SMS through Twilio, Vonage, and Plivo, Push via Firebase Cloud Messaging, Expo, and APNS, and Chat through Slack, Discord, Microsoft Teams, and WhatsApp. The embeddable React Inbox component delivers real-time WebSocket-powered notifications with read/unread states, action buttons, and user-controlled preferences directly inside your application. The workflow engine orchestrates complex multi-channel delivery with conditional branching, time delays, digest batching to reduce notification fatigue, throttle controls, and per-step channel fallbacks. Novu Connect introduces Agent Communication Infrastructure enabling AI agents built with LangChain, Claude, or custom frameworks to hold threaded two-way conversations with humans across any supported channel through a single conversation model with identity resolution and credential management. The backend runs on Node.js with NestJS, MongoDB for notification storage, Redis with BullMQ for job queuing, and Socket.io for real-time delivery. Self-hosting deploys via Docker Compose with the dashboard accessible at port 4200. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Hindsight
Memory that survives the session: Hindsight extracts facts, entities, and timestamps from everything an AI agent retains, then lets the agent search or reason over it later. Adding it to an existing LLM client takes a two-line wrapper, and more than 60 integrations cover Claude Code, Cursor, LangGraph, CrewAI, and n8n. Every bank also exposes its own MCP endpoint. Retain sends text, images, or files through an LLM that extracts facts, timestamps, entities, and relationships. Recall runs semantic, keyword, graph, and temporal searches in parallel, merges them with reciprocal rank fusion, reranks with a cross-encoder, and trims to a token budget. Reflect reasons across memories for questions that need synthesis, shaped by per-bank skepticism, literalism, and empathy traits. In the background, related facts consolidate into observations that keep exact quotes and proof counts and are refined, never silently overwritten. Mental models and wiki-style knowledge pages hold standing answers that rewrite themselves as the bank learns. Each user, agent, or project gets an isolated bank with no cross-bank leakage, and Memory Defense redacts or blocks secrets and PII against 45 patterns. The Control Plane adds a Recall Analyzer with traces, a memory constellation graph, and a knowledge page editor. Extraction and reasoning can run on 25+ LLM providers, including local Ollama and llama.cpp. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Coddy
A single static binary is all it takes to put a tireless AI agent on your server, ready to read your codebase, run shell commands, edit files, and report back through a web UI, a terminal console, your editor, or a Telegram chat. Start it as a server and the embedded web interface gives you streaming chat sessions with visible tool calls, session history, plan documents, a background task board, and a visual map of connected nodes. Point any OpenAI-compatible client at its HTTP API, or drive the same sessions from Zed, VS Code, or Obsidian through the built-in ACP editor server. Delegate bounded work to subagents that run in isolated contexts with narrowed permissions, then resume them later in their own sessions if a run stops short. Teach it your project once: rules are picked up from .coddy, .cursor, .claude, and .codex folders, AGENTS.md files are honored as work proceeds, and installable skill packs turn repeatable procedures into slash commands. Schedule recurring jobs as plain Markdown files on a cron timetable, keep long-term memory across conversations, and connect external tool servers over MCP globally or per workspace. A Prometheus exporter reports provider calls, tokens, and task health for the whole fleet. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
RedAmon
An AI agent that maps your attack surface, proves what is exploitable, and drafts the code fix: RedAmon is a red team for domains you are authorized to test. A six-phase recon pipeline runs subdomain, port, HTTP, and vulnerability scanners in parallel and merges every result into a Neo4j graph you can query in plain language. The agent works from a Kali sandbox with 100+ security tools, asks for human approval before risky steps, and can deploy a Fireteam of specialist sub-agents that investigate separate angles of one target at once. CypherFix triages each finding, then a code agent opens a GitHub pull request with the patch. AI Gauntlet attacks chatbots and LLM endpoints with prompt injection and jailbreak batteries, mapped to the OWASP LLM Top 10. TrafficMind records every HTTP request the tools send, with a searchable history the agent can replay. An Insights dashboard charts findings, CVE severity, and attack chains per project, and the settings page connects OpenAI, Anthropic, OpenRouter, Bedrock, or local Ollama and vLLM models. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Executor
Connect your services once and let every AI agent use them through a single MCP endpoint, without a credential ever reaching the agent. Executor packages each integration as an app: a small TypeScript program that turns saved accounts into tools. Install one from a catalog of more than 1,800 ready-made apps, add a remote MCP server by URL, or have your agent write and deploy a new one. Agents get two tools, skills and execute, so they search for what they need and run a short program across as many apps as a task takes. Each tool decides when to ask first: always, never, or only when the input looks risky, and you approve in your agent, your MCP client, or a browser page. The dashboard shows every app with its tools, skills, workflows, schedules, webhooks, source files, and version history, and every deploy can be rolled back. Connections give each agent its own MCP URL that reaches only the apps you pick, with copy-paste setup steps for Claude, ChatGPT, Codex, and others. Organizations, groups, invitations, shared accounts, and optional single sign-on cover teams. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
ClickHouse Loom
Give every AI agent you use the same memory, so what Claude Code learns on Monday is waiting for Codex or Cursor on Tuesday. Loom is a self-hosted server that stores memories, documents, conversations, and a code index, then serves them to agents over MCP, HTTP, and Python or TypeScript SDKs. Each fact is saved under a key with history and provenance, near-duplicates merge, and contradictions land in a review queue where a person keeps one side, drops one, or allows both. Search can abstain instead of guessing, and the Ask AI panel answers with numbered citations back to the memory or document behind each claim. A library indexes PDF, docx, and pptx files, and the repository explorer finds symbols, callers, and change impact with the commit it answers from. Namespaces, roles, and tokens let teams share memory safely, an audit log records every save, delete, and grant, and erasure requests are verified. Task inboxes pass work between agents with leases and retries. One installer script starts Loom and ClickHouse, and plugins connect Claude Code and Codex. 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.
Rakazo
Rakazo lets you hire AI bots the way you would onboard teammates: each gets a job, persistent memory, its own sandboxed computer, and a schedule, and it reports back in chat while you check in from any device. Start from templates such as Sales Outbound, Inbox Manager, Talent Scout, Expense Manager, Coding Agent, or Chief of Staff, then tune the instructions, tools, and limits until the bot works the way your team works. Every bot can run on a different model, so research uses one provider while coding uses another, and a local model served by Ollama or LM Studio keeps sensitive work off third-party servers. Bots work on real computers with a browser, terminal, file system, and graphical desktop you can watch or take over, choosing shared team computers or isolated private ones. When a task outgrows one bot, it delegates pieces to peer bots or short-lived subagents and folds their results back into its own thread. Voice mode lets you dictate messages and hear spoken replies, while routines run on schedules so follow-ups, reports, and inbox sweeps happen while you are away. You decide which actions need your approval before a bot takes them, and integrations via Composio, Pipedream Connect, MCP servers, and OpenAPI documents connect bots to tools you already use. 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.
magpie
Point every coding agent and chat client at one self-hosted gateway, then choose which model answers from a single browser console. Add providers by preset or custom OpenAI-compatible URL, or sign in with the subscriptions you already pay for. Routing groups bundle several models into one name and move to the next member when a quota runs out, so the calling agent never sees the error. Create a gateway key per client, cap each with daily token or cost budgets, restrict its models, and watch usage counted separately. The Usage pages chart tokens, cost, request counts, cache hit rate, and a yearly heatmap, split by model, provider, or agent, with CSV export. A Routing view replays how each request was routed, with timing for masking, connecting, and the vendor. Masking swaps passwords, API keys, and tokens for placeholders before a prompt leaves your server and restores them in the answer. The browser page is protected by a key you set at launch. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
OpenMA
OpenMA lets you build and run AI agents with the model, harness, and sandbox you choose, on infrastructure you control, while it handles the sessions, execution, and recovery around them. Create an agent in the console, give it work, and come back later to the same session, because conversation history, streamed progress, and execution attempts are preserved rather than tied to one process that might die. Agents start with shell, file, web, and browser tools, then grow through MCP services and reusable Skills, while resources such as GitHub repositories, uploaded files, and shared memory attach to the work that needs them. Deployments run agents manually or on schedules you define, environments carry their own networking and package configuration, and model cards keep provider settings in one place instead of scattered environment files. If a sandbox fails, the session survives it; a replacement harness can rebuild context from durable events and workspace checkpoints. Operate everything through the console, the API, or the CLI, and plug in the database, queue, and object storage your deployment already trusts. 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.
PenguinHarness
PenguinHarness turns a one-sentence description of the agent you need into a working app, then lets other agents tune it round after round until its benchmark score climbs. Nine self-improvement algorithms ship as plugins, including OPRO, APE, ACE, and AWM, and every candidate becomes a snapshotted version with its score recorded in an Evaluation Center. Seventeen ready-made benchmarks cover data analysis, repository fixes, workflow automation, and terminal science tasks. Agents read and act on web pages through a built-in browser panel or your own Chrome, and each agent can grow a hot-reloaded workbench page like a contract review desk with uploads and downloadable reports. Add a machine over ssh and conversations in that workspace run there, inside the same window. A Cost Center breaks down spend per run, trajectories show every turn on an execution timeline, and replies can carry forms, charts, and live widgets. An Agent API streams runs to outside apps, and agents can be reached from Feishu, QQ, Telegram, or WeChat. 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.
Bernstein
When an agent breaks production at 3 a.m., "the model did it" is no incident report: Bernstein decides in advance what your AI agents may touch, then keeps receipts proving what they did. You write policy as code: which roles may do what, what needs human approval, and what must be recorded before a task counts as done. A deterministic scheduler with no model in the coordination loop turns each goal into parallel tasks, so replaying yesterday's plan reproduces yesterday's task graph exactly. Every coding task runs in its own git worktree behind merge gates; nothing lands on your main branch until tests, lint, and type checks pass. It drives the CLI agents you already use (Claude Code, Codex, Gemini CLI, Aider, Goose, dozens more), mixing cheap local models with heavy cloud ones in one run. Approvals can demand a hardware key touch, and role fences hold whatever an agent tries inside its task. The journal, lineage record, and optional tamper-evident audit chain capture every step, sealed into an Ed25519-signed receipt verifiable offline from the file alone. Non-code work gets the same treatment: an artifact contract names the deliverable (report, dataset, audit evidence pack), and the task completes on a signed receipt instead of a commit. You supervise from the browser dashboard or terminal UI, watching tasks, agents, approvals, and costs live. 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.
Alook
Alook gives the coding agent on your machine a seat in the team chat, with its own handle, inbox, and room memberships, so colleagues can ask it for work the same way they would message a person. You create servers, channels, and direct messages where people and agents are both first-class members, and an agent keeps one persistent identity across every room it joins. Claude Code, Codex, Cursor, OpenCode, Pi, Grok, and Antigravity connect through a small daemon, so the agent keeps running on your own hardware while the room stays reachable from anywhere. Because the agent holds memory across rooms, it can pick up a thread in one channel, report progress in another, and move tasks forward without you relaying messages by hand. Inviting a teammate takes a link; they join the room and talk with your agent directly instead of waiting for you to copy answers back and forth. The same room follows you from desktop to phone, so a question asked at your desk is still there, with the agent's reply, when you check it later. Self-hosting runs the web app, realtime WebSocket service, email worker, and wake-up queue on your own server with a local database, started with a single npx command. 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.
OneCLI
OneCLI gives every employee a sandboxed personal AI agent that works across company tools without ever seeing the real credentials it uses. Create an agent for each person, grant only the services that person may touch, and chat with it from its dashboard page or directly inside Slack channels and direct messages. The agent keeps its own filesystem, shell, memory, and skills between conversations, so follow up messages can redirect work in progress instead of starting over. Schedules wake agents at the right time to run recurring checks, reports, and follow ups without a person remembering to prompt them. When an agent calls Gmail, Salesforce, HubSpot, GitHub, Sentry, or another connected service, the gateway checks the request against team policy and injects the stored credential only for that call. Sensitive actions can pause for a deterministic human approval in chat, so sending an email, deleting a ticket, or changing a deal stage waits for your yes. Shared team connections let administrators connect a service once and grant it per agent, while memory remains readable and editable by the owner at any time. 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.
LiteLLM Agent Control Plane
Agents scattered across vendor consoles become one tidy workspace in LiteLLM Agent Control Plane, which puts Claude Managed Agents, Cursor, OpenCode, OpenClaw, Deep Agents, and Hermes behind a single login, one API, and one dashboard. Describe an agent in plain language or start from a template for a deep researcher, an inbox triager, or a security reviewer, then tick the tools, skills, and MCP integrations it may use and pick the runtime that executes it. Chat happens inline with streamed output, sessions persist across runs, and a memory layer carries context from one session to the next. Routines launch agents on a cron schedule, and each scheduled run lands in the history as its own session. An MCP proxy injects upstream credentials at the wire, so agents never hold real secrets, and provider keys sit encrypted in a vault. Colleagues can reach the same agents from Slack, Microsoft Teams, Google Chat, or a webhook, and developers can drive everything through a REST API with a live try-it reference. Administrators issue per-user virtual keys with budgets, model limits, and expiry dates, and an inbox gathers anything that needs a human decision. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Buzz
Buzz delivers the first production workspace where humans and AI agents operate as cryptographically equal team members on a self-hosted Nostr relay. The Rust-based backend stores every message, code patch, CI result, review comment, and workflow step as a signed Nostr event in a unified PostgreSQL-backed event log with Redis pub/sub for real-time presence and S3/MinIO for media storage. The integrated Git forge implements NIP-34, turning feature branches into dedicated channels where patches, reviews, and merge decisions live alongside the discussion that produced them — eliminating the split between chat tools and code hosts. Through the open Agent Client Protocol, Buzz natively supports Goose, Anthropic Claude Code, and OpenAI Codex as first-class channel members with scoped permissions, their own audit trails, and the ability to create patches, run workflows, and orchestrate multi-step automations via YAML-defined triggers including message events, reactions, schedules, and webhooks. The Tauri-based desktop client runs on macOS, Windows, and Linux, while buzz-cli provides agent-first JSON I/O for headless automation. Deploy via Docker Compose with the production bundle in deploy/compose/, Railway one-click, or build from source requiring Rust 1.88+, Node 24+, and pnpm. Multi-community mode scopes tenant data by domain with NIP-42 Schnorr authentication, rate limiting, and hash-chain audit logging. 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.