Plugins

Runtime hot-swap for DeepSeek Harness: hot enable/disable/restart Cordis plugins and auto hot-mount bundles from the Web GUI — no dsh restart.

Web access for DeepSeek Harness — multi-provider search/fetch on ctx.web, plus GitHub/PDF/YouTube extraction.

DSH balance and quota card showing up to 3 recently used providers. Supports DeepSeek, Moonshot, Kimi For Coding, OpenAI Codex, and OpenCode Go, with auto-discovery, quota windows, reset times, color-coded status, dragging, and automatic refresh.

Hash-anchored read / edit / batch_edit / undo_last_edit tools: every line gets a unique 3-character content hash, edits target hashes instead of line numbers, and served-state verification rejects stale ranges with fresh anchors.

把 git 状态与仓库规范动态注入 system prompt：分支/脏文件数/最近提交 + 仓库规范指引（官方 system-prompt section/context/variable 缝隙插件）

Bocha-backed WebSearchProvider and installable profile bundle for DeepSeek Harness

套餐用量角标：在 Web 对话框右下角显示 5小时/周限/月限 三个用量窗口（支持 OpenCode Go、GLM Coding Plan 双渠道：Z.AI 国际版 / 智谱，以及 Kimi Code 会员套餐额度，月度会员额度可经 kimi-auth Cookie 增强），并在「设置 → 插件」提供各渠道独立的启用开关与 API Key 配置。

A read-only neuroscience data preview plugin for DeepSeek Harness (DSH)

DeepSeek Harness plugin for importing Codex App Server conversation history

AnySearch-powered real-time web and vertical search provider for DeepSeek Harness.

Japanese (日本語) interface for DeepSeek Harness — a standard DSH client plugin that adds 日本語 as a selectable UI language alongside 中文 and English, with Japanese system fonts.

Agent evaluation platform: benchmark YAML, headless run orchestration, trace-based metrics, and run reports

DeepTutor tutoring bridge: deep explanations, self-test questions, learning-path planning, personal knowledge-base search (RAG), and note archiving via deeptutor_run / deeptutor_kb / deeptutor_note tools.

Provider discovery, matching, health checks, and pre-output failover for DeepSeek Harness.

Brave Search API-backed search provider for the DeepSeek Harness web capability seam (ctx.web). One search is one HTTP request - no model turn.

Local AI code-review gate: review a workspace/branch/commit diff via the managed ocr.js, returning structured findings with token stats; gate mode fails on findings.

Per-model token cost tracking with configurable cache-hit/miss, output and peak-window pricing, a live session cost bar, and unconfigured-model flags.

Take control of your tmux panes: list/send-keys/capture, run long jobs in a pane with watch mode, and approval-gated destructive commands.

Per-process tool-call statistics: a `tool_stats` tool reporting per-tool call counts, error counts, and average durations.

Manual approval mode ("Manual Mode" / "Ask Mode").

Agent-tree token budget management.

Puts the agent on the Muretai network: its own identity, invite-based introductions, signed end-to-end encrypted messaging with agents owned by other people, and an inbound-mail wake that lets it reply on its own.

IM webhook and local notifications on turn completion, errors, or approval (Feishu/WeCom/DingTalk/Slack/Discord/custom).

Shared multi-agent task board (create/claim/transition/query) over a Cordis service key.