
Interface gráfica desktop para o prime-agent: streaming, ferramentas, árvore de subagentes RLM, explorador de arquivos e execução por SSH.
Prime Desk is a desktop GUI for prime-agent. It is a desktop client that talks to the official binary over RPC and only renders; context, compaction, skills, subagents, and the IPython kernel remain inside prime-agent. The interface provides streaming text, reasoning, and tool calls in real time, with tool cards, collapsible thinking blocks, and markdown rendering. It also includes model switching with per-Mtok cost and a reasoning indicator, seven thinking levels, searchable sessions, auto titles, folders, chat actions, an agent tree, live subagent watching, file explorer, editor, usage panel, context bar, remote execution over SSH, HUD, control features, command palette, image attachments, multiple languages, and resizable panels. Requirements include Node.js 20+, prime-agent installed on PATH and authenticated, and Linux validated on Ubuntu 24.04; macOS and Windows are untested. Prime Desk has no account of its own and uses credentials already held by prime-agent. It can authenticate via subscription (/login) for Claude Pro/Max, ChatGPT Plus/Pro (Codex), GitHub Copilot, or via API key stored in ~/.prime/agent/auth.json. The usage panel shows accumulated cost, and the site warns that subscriptions used through a third-party harness are billed per token as extra usage.
Teams use Prime Desk to run a desktop client that renders prime-agent output over RPC.
Teams use Prime Desk to stream text, reasoning, and tool calls in real time.
Teams use Prime Desk to inspect the live RLM parent→child agent hierarchy.
Teams use Prime Desk to run agent commands over SSH using the official prime-agent extension.
Teams use Prime Desk to browse and edit working-directory files with syntax highlighting.
Teams use Prime Desk to monitor sessions, messages, tokens, cost, and activity.
Teams use Prime Desk to switch models on the fly with per-Mtok cost and a reasoning indicator.
Teams use Prime Desk to attach images for multimodal models.
Lynote combines AI-draft rewriting and AI-text detection with note-taking, transcription, video summaries, and study tools.
LMSYS Chatbot Arena is a crowdsourced open platform for LLM evals. Collected over 1,000,000 human pairwise comparisons to rank LLMs with the Bradley-Terry model and display the model ratings in Elo-scale.
Zhipu Qingyan is a GLM-based AI assistant whose official description emphasizes understanding goals, breaking down tasks, and using tools.
字节跳动AI助手