When an AI agent breaks in production, Tracely freezes that exact run into a replayable regression test and blocks the pull request that would ship the bug again. Span soup hides the failure. OTLP/HTTP traces group into conversation threads, and a waterfall marks the failing span in red with its input and output. Nobody knows if an answer was good. Evaluators sit as columns on the trace table, mixing model-free checks with LLM judges at conversation, run, or span level. Thirty-one broken runs, one issue. Clustering turns them into a single counted issue and can suggest the evaluator that would have caught it. A fix nobody can prove. One click promotes a failing trace into a hermetic case with recorded fixtures and a fail-to-pass contract, and scenarios send a red-team model after an adversarial goal. Model bills on every CI run. The suite replays offline with no API keys and no model spend, and tracely gate posts a commit status, comments on the pull request, and blocks regressions. No warning when it breaks again. Alert flows drawn on a canvas react to failed gates, broken conversations, unseen failure modes, or rising failure rates by messaging Slack, sending email, calling a webhook, or prompting an LLM. Coding agents can read traces and write evaluators through an MCP endpoint. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

PanWatch's three scheduled AI agents plan the trading day before the bell, watch the session, and review it after the close for your China A-share, Hong Kong, and U.S. holdings. Day one begins with the portfolio: several brokerage accounts, holdings and profit and loss, and search for stocks or listed ETFs by code or name. The pre-market agent folds overnight moves, news, and technical structure into that plan and skips exchange holidays. Through the session, the intraday monitor watches for unusual moves and technical signals. Price alerts combine price, percentage change, turnover, and volume ratio with AND/OR logic, cooldowns, and expiry times. Each rule picks its own channel from Telegram, WeCom, DingTalk, Feishu, Bark, or a webhook. Spot something odd? The brain icon next to a holding starts TradingAgents, where four analyst roles feed a bull and bear debate, a risk review, and a portfolio-manager decision, with the reasoning trail in the app and your notifications. The opportunities screen ranks AI-scored ideas with entry levels and targets, and paper trading simulates signal-based entries and exits and tracks equity and performance. At day's end, the daily report reviews the session and prepares the next trading day's plan. OpenAI-compatible providers or local Ollama models power the analysis, and the PWA installs on a phone in English or Simplified Chinese. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Autonomous AI coding agents work side-by-side with software engineers inside Its a Plan, a keyboard-driven project tracking and sprint planning workspace designed for rapid software delivery. Product teams organize development cycles through customizable Kanban boards, prioritized backlogs, nested sub-issues, and interactive roadmap timelines designed for rapid keyboard navigation. Engineers can assign work directly to AI agents or mention them inside discussion threads to trigger automated code fixes, documentation updates, and test suite generation. Integrated runners connect to popular coding models via standard protocol interfaces, executing assigned tasks in isolated environments before reporting pull requests and results back to the issue thread. Administrators manage team permissions through role-based access controls, authenticate users with passwordless passkeys or identity providers, and broadcast instant notifications via Telegram and transactional email. The platform dispatches configurable webhooks across development tools, enabling seamless bi-directional status synchronization with external version control platforms and continuous integration pipelines. Custom labels, milestone estimates, and progress metrics provide engineering managers with immediate visibility into sprint velocity and delivery bottlenecks. 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.

salt.md is an open-source, self-hosted workspace where your team and its AI agents work on the same documents and databases, under the same permissions and with the same history. A slash-menu block editor handles checklists, tables, callouts, columns, code, diagrams, and embeds. Page links, backlinks, and a graph view show how pages connect. Any page can become a collection with typed properties like select, date, person, relation, rollup, and formula, shown as a table, board, list, gallery, calendar, timeline, or form. Ctrl+K searches titles, body text, row values, and the text inside uploaded PDFs. Two people can type in the same paragraph at once with live cursors. Comments, page history, and an activity log record who changed what. Publish a page as a read-only link with an optional password and expiry, or hand people without accounts a form that adds rows to a collection. Agents connect over MCP with 33 tools for searching, writing, building databases, and commenting, using tokens scoped to one workspace or read-only. Signed webhooks, iCalendar feeds, and Google or Microsoft sign-in link it to the rest of your stack. Import Markdown, ZIP archives, or a Notion export, and export pages as Markdown, HTML, or PDF. Use it for handbooks, meeting notes, project boards, and a knowledge base your assistants can read. 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.

Support teams handling website live chat, email queues, and social messaging channels gain a unified, AI-assisted helpdesk in Bytedesk. Incoming conversations from web widgets, mobile SDKs, WhatsApp, WeChat, and email flow into a single agent workbench equipped with skill-group routing, VIP priority rules, and automatic human handoff. Customer-facing AI agents connect to Ollama, DeepSeek, OpenAI, Qwen, and Zhipu models to answer recurring inquiries using retrieval-augmented generation (RAG) over internal knowledge bases and FAQ libraries, complete with function calling and Model Context Protocol (MCP) tool execution. Complex escalations convert into SLA-tracked tickets with custom approval workflows, or transition directly into browser-based WebRTC video calls, screen sharing sessions, and FreeSwitch voice call center queues. Supervisors track real-time queue depth, bot resolution rates, token usage, automated AI quality scores, and customer satisfaction surveys from the analytics dashboard. Native client SDKs for Web, iOS, Android, Flutter, UniApp, WordPress, and WooCommerce make embedding customizable chat widgets straightforward across any storefront. 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.