TraceRoot
An agent run can finish without errors and still give the wrong answer, so TraceRoot traces every model and tool call, flags the production runs that missed the mark, and helps your coding agent ship verified fixes. Each run becomes a browsable call tree with inputs, outputs, latency, and cost, searchable by session or filter. Instrument with the Python or TypeScript SDK, the agent-native CLI, or any OpenTelemetry exporter, with ready integrations for LangChain, LangGraph, OpenAI Agents SDK, CrewAI, Vercel AI SDK, Mastra, and Pydantic AI. Detectors use LLM judges to score live traffic against the behavior you define and can post alerts to Slack. Judge outputs cluster into signals, then a background coding agent reads the traces and your connected GitHub code in a sandbox and recommends changes down to the commit, file, and line. Datasets, test cases, and scored eval runs compare versions so regressions surface before a release, and dashboards of widgets chart your trace data. The CLI hands your coding agent, such as Claude Code or Codex, the traces, signals, detectors, and evals to work from. Self-hosting keeps all of that trace data on infrastructure you control. 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 (core edition).
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.
MaxKB
Support teams, IT departments, and researchers can put their own documents behind chatbots and AI agents built with MaxKB, an open-source platform that grounds answers in private knowledge. A support lead starts by uploading Markdown, DOCX, PDF, TXT, HTML, XLSX, XLS, or CSV files, or by syncing a web site, and MaxKB splits, vectorizes, and indexes the content on its own. An IT department can place the finished agent in any third-party site or internal tool with no coding, as a full-screen page, a mobile view, or a floating chat window. Researchers get replies that show, and let them download, the knowledge sources behind each answer. Simple agents pick tools, skills, and MCP connections from the question itself, while advanced agents are drawn as workflows with branching, multi-path recall, loops, and variables. Custom scripts, which AI can help write, plus schedules and event triggers extend what an agent can do. The model manager registers local models such as DeepSeek, Llama, and Qwen beside providers such as OpenAI, Claude, and Gemini, across language, embedding, reranking, vision, speech, and image generation. Conversation logs feed analysis for operations. Dozens of ready-made agent templates, knowledge base templates, tools, and skills in the template center and tool store shorten the first build. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GPL-3.0 licensed.
AgenTerm
AgenTerm lets your AI agent work inside a remote computer's terminal with the owner's permission, through a relay server you host yourself. Your agent connects to the server over MCP and creates an invitation, single-use and expiring in fifteen minutes by default, or reusable across several devices; you privately share the generated connection command with the person who needs help. That person runs a small client on macOS, Linux, or Windows, reads what access means, and authorizes it; the client then dials out to your server, so the target computer needs no SSH service and no open inbound port. The agent can then run one-off commands with a chosen working directory, environment overrides, and timeouts, receiving stdout, stderr, and exit status, or hold an interactive session open, reading output as it arrives and typing answers to prompts. The person at the keyboard watches every command and its output in the client and can disconnect at any moment, and the client reconnects by itself after a network drop. Use it to troubleshoot a colleague's machine without swapping logs, help family through a terminal problem, or drive a development box with no agent or model account of its own. The relay itself is a single Go binary guarded by a management key, deployed with Docker Compose or systemd, with in-memory sessions and a health endpoint. 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.
AgentDock
AgentDock turns the computers and servers you already run into machines an AI assistant can operate for you through one secure MCP endpoint. Connect ChatGPT, Claude, Codex, or any other MCP client and the agent works in the real environment: reading and writing files, searching code, running shell commands, and driving Git where the project actually lives. Long jobs survive interruptions because task state and checkpoints persist on the node, ready to resume where they stopped. Browser work is covered as well, with web pages driven through a native Chrome DevTools connection that reuses your logged-in sessions, captures screenshots, and can even operate macOS desktop applications. Capabilities grow through installable Skills and dynamic MCP servers, so a node picks up new tools without a rebuild. Run a single node on a laptop, or spread nodes across VPS hosts and containers and pair them with the optional NexusDock console for one fleet view, shared Recall memory, reusable workflow templates, and a single MCP address for every machine. Access stays locked down with Bearer tokens or OAuth sign-in, loopback-only defaults, and optional Cloudflare Tunnel or HTTPS exposure when a remote client needs a way in. 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.
LangWatch
Every call your AI agents make becomes a searchable trace in LangWatch, and the same platform tests and scores those agents before a new prompt or model reaches users. The Trace Explorer filters by origin, user, model, cost, and evaluator, reads each run as a waterfall, flame graph, topology, or sequence diagram, and takes traces from the Python, TypeScript, and Go SDKs or any OpenTelemetry source. Agent Testing sends simulated users, in text or beta voice calls, through your agent while a judge passes or fails every criterion with its reasoning, adds multi-turn red teaming, and compares two agents side by side. The Experiments workbench scores prompts and agents on datasets with LLM judges, exact match, and RAG checks, then compares runs on cost, latency, and pass rate, while online evaluations monitor live traffic and guardrails block unsafe replies. Prompts keep version history and staging or production tags, Workflows wire prompts, Python code, and HTTP calls on a visual canvas, and annotation queues route traces to human reviewers. Alerts fire when a graph crosses a threshold, automations post to Slack, email, or webhooks, and scheduled reports go out by email or Slack. Claude Code, Codex, Copilot, Gemini CLI, and opencode report session cost, pull request cost, and terminal replays, and an MCP server exposes trace search and prompt management to coding assistants. 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.
Leon
Leon is a personal AI assistant that lives on your own server, remembers what matters to you, and finishes real tasks with tools instead of just chatting. Ask for something in the browser web app and Leon plans the work, runs tools in isolated workers, checks the results, recovers from failures, and writes a short summary. Agent mode handles open-ended requests, controlled mode follows predictable native skills, and smart mode picks the better route for each task. Layered memory keeps durable facts and preferences, daily summaries, and recent discussion context, and Leon can search earlier conversations when exact details matter. Answers can arrive as interactive charts with region toggles, forms, and calculators instead of walls of text, and generated files become downloadable artifacts. Setup cards guide you through connecting services such as Spotify, owner-installed MCP servers add more tools, and a browser tool and computer use handle web pages and desktop applications. Pick models per profile, from a local llama.cpp model to OpenAI, Anthropic, DeepSeek, Groq, or OpenRouter. Profiles keep config, secrets, sessions, and memory separate, so one server can serve several people, with optional token authentication for remote clients. Workspaces group conversations under a shared brief, and Leon Satellite lets a remote Leon use enabled tools on your own device. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
ZeroClaw
ZeroClaw puts one personal AI agent on every messaging app you already use, answering from your own server with your own model keys. A single instance binds Discord, Telegram, Matrix, Slack, WhatsApp, Signal, email, and webhooks at once, so you can start a conversation on your phone and finish it at your desk without losing context. You choose the brain behind it: Anthropic, OpenAI, a local Ollama model, or any OpenAI-compatible endpoint, with fallback chains that keep the agent answering when one provider has a bad day. The agent acts through tools that run shell commands, drive a browser, call HTTP APIs, and talk to custom MCP servers you plug in. A web dashboard gives you chat, a memory browser, a configuration editor, cron job management, and a tool inspector from any browser. Long-term memory lives in SQLite on your disk with embedding-based recall, so past conversations stay searchable and private. Supervised autonomy is the default posture: medium-risk actions pause for your approval, high-risk actions are blocked outright, and operating system sandboxes plus cryptographic receipts keep every tool call bounded and auditable. Event-driven standard operating procedures react to MQTT messages, webhooks, schedules, and hardware events, and named agents let different personalities guard different channels. Registered as a system service, it stays online around the clock. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT or Apache-2.0 licensed.
ChatDev
Built around a drag-and-configure canvas of nodes and edges, ChatDev runs teams of AI agents as workflows you can watch, steer, and reuse. Nodes get a role, a model, and a system prompt from a plain form, and edges decide who speaks next. Agent nodes call OpenAI, Gemini, or any OpenAI-compatible endpoint. Human nodes pause the run and wait for your notes or attached files. Edges can branch on keywords or custom functions, extract values with regex, and loop through review-and-revise cycles guarded by loop counters and timers. Map and Tree edges fan one message out to parallel instances, and Tree mode reduces the results into a single answer. Agents can use simple, file, blackboard, or Mem0 memory, reflection thinking, built-in file, Python environment, and deep research tools, Agent Skills, and MCP servers over HTTP or stdio. Dozens of bundled workflows cover software teams, game development, data visualization, and deep research. The Launch page shows each node with a pixel-art avatar, a chat view, file uploads, and downloadable logs and artifacts. A Laboratory page runs one workflow over every row of a CSV or Excel file in parallel and tracks the success rate. The Python SDK runs YAML workflows from scripts. An illustrated tutorial and an English or Chinese interface get a first workflow running in minutes. 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.
AgentVerse OS
AgentVerse OS turns a single Ubuntu server into a personal cloud with a real desktop in the browser, built for a developer and their AI agents. You work in a windowed desktop with a taskbar, start menu, and widgets for system monitoring, clock, news, weather, and projects, and the same desktop opens from a laptop, tablet, or phone while everything stays on the server. Each project gets an isolated Incus container with Docker inside, its own network and gate, plus VS Code, a terminal, files, and git in the browser. The Agents app runs Claude Code, Codex, Gemini CLI, pi, and Antigravity inside the workspace, with permission prompts you approve from any device, scheduled routines, model menus, conversation search and export, and a ledger of what each session spent. When you close your laptop the agents keep working, and you pick the conversation up later exactly where it stopped. A built-in store merges the Runtipi, Coolify, and Umbrel catalogs into about 945 installable apps with guided setup, credentials shown in the app card, snapshots, and rollback. Projects request capabilities such as storage.s3, llm, or notify instead of hardcoding addresses, scheduled ZFS snapshots feed a restic backup repository, and access arrives only through your Tailscale network with real certificates. 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.
Edict
Edict is a multi-agent orchestrator that runs every AI request through a court-style chain: a crown prince agent triages it, a planning department drafts the work, a review department can veto the plan, and specialist ministries execute what survives. Casual chat gets a direct reply, while real commands become kanban tickets whose titles are scrubbed of file paths and metadata. A vetoed plan returns for rework; an approved one is dispatched in parallel to ministries for finance, documentation, engineering, compliance, infrastructure, and HR. Commands arrive through Feishu, Telegram, or Signal channels configured in OpenClaw. The dashboard shows cards by state with heartbeat badges and stop, cancel, and resume controls, plus a department monitor and a memorial archive that turns finished work into five-phase timelines you can copy as Markdown. Nine edict templates add parameter forms and cost estimates, an officials panel ranks agents by merit score while tracking token spend, and per-agent model switching restarts the gateway in about five seconds. Agent thinking, tool calls, and results stream live. A permission matrix limits who may message whom, illegal state jumps are rejected, and every transition lands in an audit log. Worked examples cover competitive analysis, code review, and weekly reports. Dashboard login protects the console, and a systemd unit keeps it running across reboots. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
AI-Infra-Guard
From MCP servers and agent skills to Ollama endpoints and LLM guardrails, AI-Infra-Guard probes the whole AI stack for security holes from one chat-style web console. Infra Scan fingerprints 100+ AI frameworks such as Ollama, ComfyUI, vLLM, Gradio, and Triton, then matches versions against 2,000+ known CVEs for single targets or uploaded batch lists, with port scanning for bare IPs. MCP Scan audits a server from a zip file, a GitHub repository, or a live SSE or streamable HTTP address, hunting for tool poisoning, rug pulls, data theft, indirect prompt injection, and plaintext keys. Skill Scan puts agent skills through the same three stages of information collection, code audit, and vulnerability review. Agent Scan tests Dify, Coze, and HTTP chat agents against the OWASP Top 10 for agentic apps, and Jailbreak Evaluation fires curated datasets and attack methods at any OpenAI-compatible model, then exports the failed cases. Reports give a score, a security level, evidence, an attack path, and fixes, while progress streams live as the scanning agents work. Plugin management edits fingerprint rules, vulnerability entries, evaluation datasets, and MCP plugins right in the browser. You bring your own LLM key for the scanning agents. There is no built-in login, so keep it on a private network or behind a reverse proxy. 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.
Agent Swarm
Agent Swarm gives a company one lead AI agent and a crew of coding workers that take requests from Slack, GitHub, Linear, or email and hand back pull requests while people only approve. The lead reads each request, plans it, and delegates to workers that run Claude Code, Codex, pi, opencode, Cursor, or Amp in isolated containers with a full development environment. Every agent keeps a persistent identity, and what one learns goes into shared memory with hybrid keyword and vector search, so the next task starts with context. Schedules run recurring jobs on cron or interval timers, and a DAG workflow engine adds triggers, conditions, per-step retries, and approval nodes that pause until someone answers in the dashboard or in Slack. The dashboard lists tasks, sessions, approvals, agents, people, workflows, scripts, skills, MCP servers, memory, and per-user usage costs. Its Apps feature builds versioned, role-restricted pages such as an incident board or a launch tracker, each with history and rollback. A configuration page tunes steering, memory, and integrations without editing environment files, and official templates supply ready-made agents for coding, review, research, QA, and content work. Workers can expose HTTP services, call approved third-party routes, and pay for x402 APIs in USDC. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Versus Incident
Versus Incident is a self-hosted AI SRE agent and incident manager that learns what normal looks like in your logs, metrics, traces, and Kubernetes clusters, then opens a triaged incident the moment something genuinely new appears. Detection earns trust gradually. Training mode only observes and catalogs patterns; shadow mode records what it would have alerted so you can judge its instincts; detect mode creates real incidents with an AI-written summary, a severity, and suggested next steps attached. Existing alerting does not have to move: Alertmanager, Grafana, Sentry, CloudWatch, Fluent Bit, and custom scripts post to the same webhook endpoint, and every incident, however it arrived, is rendered with your Go templates. Delivery fans out to Slack, Microsoft Teams, Telegram, Viber, email, and Lark. Silence has a deadline, because unacknowledged incidents escalate to on-call through PagerDuty, AWS Incident Manager, Opsgenie, incident.io, or ServiceNow. The Kubernetes dashboard gathers cluster health and capacity, workloads, resource relationships, recent changes, Helm releases, and GitOps status, and a DevOps chat agent in the web UI answers questions against the connected systems while responders work. The admin dashboard reviews learned patterns, shadow events, and raised incidents, so the agent's judgment stays inspectable. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Latitude
Production AI agents fail in quiet ways, and Latitude records every trace, session, and tool call so each failure can be found, grouped, and fixed. One line of telemetry from the TypeScript or Python SDK, or any OpenTelemetry exporter, sends traces from OpenAI, Anthropic, LangChain, LlamaIndex, CrewAI, and dozens of other providers and frameworks. Failing traces are clustered into signals with status, size, and trend, and flaggers and annotation queues let reviewers score outputs by hand. Evaluations are generated from those judgments and tracked with an alignment score. Semantic and exact-text search covers every trace, and saved searches, filters, and behavior groupings keep recurring questions one click away. Dashboards report cost per session, tool failure rates, and agent score across outcome, reliability, speed, and safety. Memory views show how an agent's stored memory changes over time, with diffs and the trace behind each edit. Monitors raise alerts, regression datasets replay fixes against the traces that failed, and Agent Dispatch hands the evidence to a coding agent. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
CRW
CRW gives your apps and AI agents a private web data API that scrapes pages, crawls sites, maps links, and searches the web without sending every request to a hosted scraping service. Post a URL to the scrape endpoint and receive clean Markdown, HTML, screenshots, links, or structured JSON shaped by a schema you supply. Start a crawl when a whole section matters, then poll for progress while CRW walks pages asynchronously and returns the collected content in one job. Use map first to discover the URLs on a site before spending effort extracting pages you do not need. Connect agent tools through the MCP endpoint so assistants can search, scrape, and crawl from the same self-hosted service your applications call. Add browser rendering when JavaScript-heavy pages hide content from a plain HTTP fetch, and keep simpler pages on the fast lightweight path. Protect a public deployment with Bearer token authentication, your own reverse proxy, logging, and rate limits, or keep the service inside a private network for internal tools. Run one Docker container for the core API, or use the Compose stack when you want bundled search and rendering sidecars beside it. 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.
Argus Agentic SOC
Every incident ticket in Argus, an open-source security operations center, comes with its own AI analyst, ready to explain the alert and suggest what to do next. Scheduled pipeline tasks pull alerts in from Elasticsearch and other sources, and tickets carry severity, owner, and SLA clocks that measure MTTA, MTTI, MTTC, and MTTR with color-coded buckets. Decision buttons walk each case from New through Acknowledged, Triaged, and Contained to Resolved. A War Room gathers analyst notes, system logs, files, pasted images, and extracted indicators. One click on Run AI Assistant returns a plain-language alert explanation, a suggested risk level, finished tasks, and next steps; scoped chat plus @ai and @playbook mentions keep the conversation inside the ticket. Built-in MCP tools look up CMDB assets, find similar past cases, and extract IPs, domains, and hashes, and you can register your own MCP servers and SKILL.md skills with usage monitors. A drag-and-drop workflow editor fires playbooks on new alerts, ticket changes, cron schedules, or webhooks. Detection engineers import Sigma rules from GitHub URLs or files, map fields to Splunk and Elastic ECS names, and review a publish history. 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.
Marinara Engine
Marinara Engine turns any AI model into a character chat, a roleplay stage, or a tabletop-style game run by an AI Game Master, all in one self-hosted web app. Conversation mode feels like a messenger, with group chats, audio and video calls, selfies, and characters who keep daily schedules and message you first. Roleplay mode adds a stage behind the words: character sprites, scene backgrounds, weather effects, and trackers for time, quests, stats, and mood. Game mode hands the story to a Game Master that runs NPCs, dice and skill checks, classic or tactical battles, maps, and saved sessions on rulesets such as 5e (SRD 5.1). Connect OpenAI, Anthropic, Gemini, OpenRouter, DeepSeek, or any OpenAI-compatible endpoint, run local models through Ollama, LM Studio, or KoboldCpp, or sign in with a Claude, ChatGPT, or Grok plan instead of an API key. Optional agents work beside each chat to check continuity, track the world, draw illustrations, play music, and recall long-term memory, while Professor Mari, the built-in assistant, helps with setup. Lorebooks add world facts when keywords appear, and Peek Prompt shows exactly what the model received. Import characters, chats, and lorebooks from SillyTavern, then export chats as JSONL, Markdown, or styled HTML. Visitors from the network or internet must sign in, with Basic Auth and IP allowlists for extra control. 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.