mirror of
https://github.com/Ed1s0nZ/CyberStrikeAI.git
synced 2026-04-21 10:16:32 +02:00
Add files via upload
This commit is contained in:
公明
@@ -111,13 +111,13 @@ CyberStrikeAI is an **AI-native security testing platform** built in Go. It inte
|
||||
- 📄 Large-result pagination, compression, and searchable archives
|
||||
- 🔗 Attack-chain graph, risk scoring, and step-by-step replay
|
||||
- 🔒 Password-protected web UI, audit logs, and SQLite persistence
|
||||
- 📚 Knowledge base with embedding-based vector retrieval (cosine similarity) for security expertise
|
||||
- 📚 Knowledge base (RAG) with embedding-based vector retrieval (cosine similarity), optional **Eino Compose** indexing pipeline, and configurable post-retrieval budgets / reranking hooks
|
||||
- 📁 Conversation grouping with pinning, rename, and batch management
|
||||
- 🛡️ Vulnerability management with CRUD operations, severity tracking, status workflow, and statistics
|
||||
- 📋 Batch task management: create task queues, add multiple tasks, and execute them sequentially
|
||||
- 🎭 Role-based testing: predefined security testing roles (Penetration Testing, CTF, Web App Scanning, etc.) with custom prompts and tool restrictions
|
||||
- 🧩 **Multi-agent mode (Eino DeepAgent)**: optional orchestration where a coordinator delegates work to Markdown-defined sub-agents via the `task` tool; main agent in `agents/orchestrator.md` (or `kind: orchestrator`), sub-agents under `agents/*.md`; chat mode switch when `multi_agent.enabled` is true (see [Multi-agent doc](docs/MULTI_AGENT_EINO.md))
|
||||
- 🎯 Skills system: 20+ predefined security testing skills (SQL injection, XSS, API security, etc.) that can be attached to roles or called on-demand by AI agents
|
||||
- 🧩 **Multi-agent (CloudWeGo Eino)**: alongside **single-agent ReAct** (`/api/agent-loop`), **multi mode** (`/api/multi-agent/stream`) offers **`deep`** (coordinator + `task` sub-agents), **`plan_execute`** (planner / executor / replanner), and **`supervisor`** (orchestrator + `transfer` / `exit`); chosen per request via **`orchestration`**. Markdown under `agents/`: `orchestrator.md` (Deep), `orchestrator-plan-execute.md`, `orchestrator-supervisor.md`, plus sub-agent `*.md` where applicable (see [Multi-agent doc](docs/MULTI_AGENT_EINO.md))
|
||||
- 🎯 **Skills (refactored for Eino)**: packs under `skills_dir` follow **Agent Skills** layout (`SKILL.md` + optional files); **multi-agent** sessions use the official Eino ADK **`skill`** tool for **progressive disclosure** (load by name), with optional **host filesystem / shell** via `multi_agent.eino_skills`; optional **`eino_middleware`** adds patchtoolcalls, tool_search, plantask, reduction, checkpoints, and Deep tuning—20+ sample domains (SQLi, XSS, API security, …) can still be bound to roles
|
||||
- 📱 **Chatbot**: DingTalk and Lark (Feishu) long-lived connections so you can talk to CyberStrikeAI from mobile (see [Robot / Chatbot guide](docs/robot_en.md) for setup and commands)
|
||||
- 🐚 **WebShell management**: Add and manage WebShell connections (e.g. IceSword/AntSword compatible), use a virtual terminal for command execution, a built-in file manager for file operations, and an AI assistant tab that orchestrates tests and keeps per-connection conversation history; supports PHP, ASP, ASPX, JSP and custom shell types with configurable request method and command parameter.
|
||||
|
||||
@@ -228,7 +228,7 @@ Requirements / tips:
|
||||
|
||||
### Core Workflows
|
||||
- **Conversation testing** – Natural-language prompts trigger toolchains with streaming SSE output.
|
||||
- **Single vs multi-agent** – With `multi_agent.enabled: true`, the chat UI can switch between **single** (classic ReAct loop) and **multi** (Eino DeepAgent + `task` sub-agents). Multi mode uses `/api/multi-agent/stream`; tools are bridged from the same MCP stack as single-agent.
|
||||
- **Single vs multi-agent** – With `multi_agent.enabled: true`, the chat UI can switch between **single** (classic **ReAct** loop, `/api/agent-loop/stream`) and **multi** (`/api/multi-agent/stream`). Multi mode keeps **`deep`** as the baseline coordinator + **`task`** sub-agents, and adds **`plan_execute`** and **`supervisor`** orchestrations via the request body **`orchestration`** field. MCP tools are bridged the same way as single-agent.
|
||||
- **Role-based testing** – Select from predefined security testing roles (Penetration Testing, CTF, Web App Scanning, API Security Testing, etc.) to customize AI behavior and tool availability. Each role applies custom system prompts and can restrict available tools for focused testing scenarios.
|
||||
- **Tool monitor** – Inspect running jobs, execution logs, and large-result attachments.
|
||||
- **History & audit** – Every conversation and tool invocation is stored in SQLite with replay.
|
||||
@@ -250,7 +250,7 @@ Requirements / tips:
|
||||
- **Predefined roles** – System includes 12+ predefined security testing roles (Penetration Testing, CTF, Web App Scanning, API Security Testing, Binary Analysis, Cloud Security Audit, etc.) in the `roles/` directory.
|
||||
- **Custom prompts** – Each role can define a `user_prompt` that prepends to user messages, guiding the AI to adopt specialized testing methodologies and focus areas.
|
||||
- **Tool restrictions** – Roles can specify a `tools` list to limit available tools, ensuring focused testing workflows (e.g., CTF role restricts to CTF-specific utilities).
|
||||
- **Skills integration** – Roles can attach security testing skills. Skill ids are hinted in the system prompt; in **multi-agent (DeepAgent)** sessions the Eino ADK **`skill`** tool loads package content on demand (the single-agent path has no skill middleware yet).
|
||||
- **Skills integration** – Roles can attach security testing skills. Skill ids are hinted in the system prompt; in **multi-agent** sessions the Eino ADK **`skill`** tool loads package content **on demand** (progressive disclosure). **`multi_agent.eino_skills`** toggles the middleware, tool name override, and optional **read_file / glob / grep / write / edit / execute** on the host (**Deep / Supervisor** main and sub-agents when enabled; **plan_execute** executor has no custom skill middleware—see docs). Single-agent ReAct does not mount this Eino skill stack today.
|
||||
- **Easy role creation** – Create custom roles by adding YAML files to the `roles/` directory. Each role defines `name`, `description`, `user_prompt`, `icon`, `tools`, `skills`, and `enabled` fields.
|
||||
- **Web UI integration** – Select roles from a dropdown in the chat interface. Role selection affects both AI behavior and available tool suggestions.
|
||||
|
||||
@@ -271,20 +271,24 @@ Requirements / tips:
|
||||
```
|
||||
2. Restart the server or reload configuration; the role appears in the role selector dropdown.
|
||||
|
||||
### Multi-Agent Mode (Eino DeepAgent)
|
||||
- **What it is** – An optional second execution path based on CloudWeGo **Eino** `adk/prebuilt/deep`: a **coordinator** (main agent) calls a **`task`** tool to run ephemeral **sub-agents**, each with its own model loop and tool set derived from the current role.
|
||||
- **Markdown agents** – Under `agents_dir` (default `agents/`, relative to `config.yaml`), define:
|
||||
- **Orchestrator**: file name `orchestrator.md` *or* any `.md` with front matter `kind: orchestrator` (only **one** per directory). Sets Deep agent name/id, description, and optional full system prompt (body); if the body is empty, `multi_agent.orchestrator_instruction` and then Eino defaults apply.
|
||||
- **Sub-agents**: other `*.md` files (YAML front matter + body as instruction). They are **not** used as `task` targets if classified as orchestrator.
|
||||
- **Management** – Web UI: **Agents → Agent management** for CRUD on Markdown agents; API prefix `/api/multi-agent/markdown-agents`.
|
||||
- **Config** – `multi_agent` block in `config.yaml`: `enabled`, `default_mode` (`single` | `multi`), `robot_use_multi_agent`, `batch_use_multi_agent`, `max_iteration`, `orchestrator_instruction`, optional YAML `sub_agents` merged with disk (same `id` → Markdown wins).
|
||||
- **Details** – Streaming events, robots, batch queue, and troubleshooting: **[docs/MULTI_AGENT_EINO.md](docs/MULTI_AGENT_EINO.md)**.
|
||||
### Multi-Agent Mode (Eino: Deep, Plan-Execute, Supervisor)
|
||||
- **What it is** – An optional execution path beside **single-agent ReAct**, built on CloudWeGo **Eino** `adk/prebuilt`: **`deep`** — coordinator + **`task`** sub-agents; **`plan_execute`** — planner / executor / replanner loop (no YAML/Markdown sub-agent list); **`supervisor`** — orchestrator with **`transfer`** and **`exit`** over Markdown-defined specialists. The client sends **`orchestration`**: `deep` | `plan_execute` | `supervisor` (default `deep`).
|
||||
- **Markdown agents** – Under `agents_dir` (default `agents/`):
|
||||
- **Deep orchestrator**: `orchestrator.md` *or* one `.md` with `kind: orchestrator`. Body or `multi_agent.orchestrator_instruction`, then Eino defaults.
|
||||
- **Plan-Execute orchestrator**: fixed name **`orchestrator-plan-execute.md`** (plus optional `orchestrator_instruction_plan_execute` in YAML).
|
||||
- **Supervisor orchestrator**: fixed name **`orchestrator-supervisor.md`** (plus optional `orchestrator_instruction_supervisor`); requires at least one sub-agent.
|
||||
- **Sub-agents** (for **deep** / **supervisor**): other `*.md` files (YAML front matter + body). Not used as **`task`** targets if marked orchestrator-only.
|
||||
- **Management** – Web UI: **Agents → Agent management**; API `/api/multi-agent/markdown-agents`.
|
||||
- **Config** – `multi_agent` in `config.yaml`: `enabled`, `default_mode`, `robot_use_multi_agent`, `batch_use_multi_agent`, `max_iteration`, `plan_execute_loop_max_iterations`, per-mode orchestrator instruction fields, optional YAML `sub_agents` merged with disk (`id` clash → Markdown wins), **`eino_skills`**, **`eino_middleware`** (optional ADK middleware and Deep/Supervisor tuning).
|
||||
- **Details** – **[docs/MULTI_AGENT_EINO.md](docs/MULTI_AGENT_EINO.md)** (streaming, robots, batch, middleware caveats).
|
||||
|
||||
### Skills System (Agent Skills + Eino)
|
||||
- **Layout** – Each skill is a directory with **required** `SKILL.md` only ([Agent Skills](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview)): YAML front matter **only** `name` and `description`, plus Markdown body. Optional sibling files (`FORMS.md`, `REFERENCE.md`, `scripts/*`, …). **No** `SKILL.yaml` (not part of Claude or Eino specs); sections/scripts/progressive behavior are **derived at runtime** from Markdown and the filesystem.
|
||||
- **Eino** – Packages are split into `schema.Document` chunks for `FilesystemSkillsRetriever` (`skills.AsEinoRetriever()`) in compose/graphs.
|
||||
- **Runtime refactor** – **`skills_dir`** is the single root for packs. **Multi-agent** loads them through Eino’s official **`skill`** middleware (**progressive disclosure**: model calls `skill` with a pack **name** instead of receiving full SKILL text up front). Configure via **`multi_agent.eino_skills`**: `disable`, `filesystem_tools` (host read/glob/grep/write/edit/execute), `skill_tool_name`.
|
||||
- **Eino / RAG** – Packages are also split into `schema.Document` chunks for `FilesystemSkillsRetriever` (`skills.AsEinoRetriever()`) in **compose** graphs (e.g. knowledge/indexing pipelines).
|
||||
- **Skill hints in prompts** – Role-bound skill **ids** (directory names) are recommended in the system prompt; full text is not injected by default.
|
||||
- **HTTP API** – `/api/skills` listing and `depth` (`summary` | `full`), `section`, and `resource_path` remain for the web UI and ops; **model-side** skill loading uses the multi-agent **`skill`** tool, not MCP.
|
||||
- **HTTP API** – `/api/skills` listing and `depth` (`summary` | `full`), `section`, and `resource_path` remain for the web UI and ops; **model-side** skill loading in multi-agent uses the **`skill`** tool, not MCP.
|
||||
- **Optional `eino_middleware`** – e.g. `tool_search` (dynamic MCP tool list), `patch_tool_calls`, `plantask` (structured tasks; persistence defaults under a subdirectory of `skills_dir`), `reduction`, `checkpoint_dir`, Deep output key / model retries / task-tool description prefix—see `config.yaml` and `internal/config/config.go`.
|
||||
- **Shipped demo** – `skills/cyberstrike-eino-demo/`; see `skills/README.md`.
|
||||
|
||||
**Creating a skill:**
|
||||
@@ -522,7 +526,10 @@ multi_agent:
|
||||
default_mode: "single" # single | multi (UI default when multi-agent is enabled)
|
||||
robot_use_multi_agent: false
|
||||
batch_use_multi_agent: false
|
||||
orchestrator_instruction: "" # Optional; used when orchestrator.md body is empty
|
||||
orchestrator_instruction: "" # Deep; used when orchestrator.md body is empty
|
||||
# orchestrator_instruction_plan_execute / orchestrator_instruction_supervisor optional
|
||||
# eino_skills: { disable: false, filesystem_tools: true, skill_tool_name: skill }
|
||||
# eino_middleware: optional patch_tool_calls, tool_search, plantask, reduction, checkpoint_dir, ...
|
||||
```
|
||||
|
||||
### Tool Definition Example (`tools/nmap.yaml`)
|
||||
@@ -567,7 +574,7 @@ enabled: true
|
||||
|
||||
## Related documentation
|
||||
|
||||
- [Multi-agent mode (Eino)](docs/MULTI_AGENT_EINO.md): DeepAgent orchestration, `agents/*.md`, APIs, and chat/stream behavior.
|
||||
- [Multi-agent mode (Eino)](docs/MULTI_AGENT_EINO.md): **Deep**, **Plan-Execute**, **Supervisor**, `agents/*.md`, `eino_skills` / `eino_middleware`, APIs, and chat/stream behavior.
|
||||
- [Robot / Chatbot guide (DingTalk & Lark)](docs/robot_en.md): Full setup, commands, and troubleshooting for using CyberStrikeAI from DingTalk or Lark on your phone. **Follow this doc to avoid common pitfalls.**
|
||||
|
||||
## Project Layout
|
||||
|
||||
+24
-17
@@ -110,13 +110,13 @@ CyberStrikeAI 是一款 **AI 原生安全测试平台**，基于 Go 构建，集
|
||||
- 📄 大结果分页、压缩与全文检索
|
||||
- 🔗 攻击链可视化、风险打分与步骤回放
|
||||
- 🔒 Web 登录保护、审计日志、SQLite 持久化
|
||||
- 📚 知识库功能：向量检索（与 Eino Retriever 语义一致），为 AI 提供安全专业知识
|
||||
- 📚 知识库（RAG）：向量嵌入与余弦相似度检索（与 Eino `retriever.Retriever` 语义一致），可选 **Eino Compose** 索引流水线及检索后处理（预算、重排等配置项）
|
||||
- 📁 对话分组管理：支持分组创建、置顶、重命名、删除等操作
|
||||
- 🛡️ 漏洞管理功能：完整的漏洞 CRUD 操作，支持严重程度分级、状态流转、按对话/严重程度/状态过滤，以及统计看板
|
||||
- 📋 批量任务管理：创建任务队列，批量添加任务，依次顺序执行，支持任务编辑与状态跟踪
|
||||
- 🎭 角色化测试：预设安全测试角色（渗透测试、CTF、Web 应用扫描等），支持自定义提示词和工具限制
|
||||
- 🧩 **多代理模式（Eino DeepAgent）**：可选编排——协调主代理通过 `task` 调度 Markdown 定义的子代理；主代理见 `agents/orchestrator.md` 或 front matter `kind: orchestrator`，子代理为 `agents/*.md`；开启 `multi_agent.enabled` 后聊天可切换单代理/多代理（详见 [多代理说明](docs/MULTI_AGENT_EINO.md)）
|
||||
- 🎯 Skills 技能系统：20+ 预设安全测试技能（SQL 注入、XSS、API 安全等），可附加到角色或由 AI 按需调用
|
||||
- 🧩 **多代理（CloudWeGo Eino）**：在 **单代理 ReAct**（`/api/agent-loop`）之外，**多代理**（`/api/multi-agent/stream`）提供 **`deep`**（协调主代理 + `task` 子代理）、**`plan_execute`**（规划 / 执行 / 重规划）、**`supervisor`**（主代理 `transfer` / `exit` 监督子代理）；由请求体 **`orchestration`** 选择。`agents/` 下分模式主代理：`orchestrator.md`（Deep）、`orchestrator-plan-execute.md`、`orchestrator-supervisor.md`，及适用的子代理 `*.md`（详见 [多代理说明](docs/MULTI_AGENT_EINO.md)）
|
||||
- 🎯 **Skills（面向 Eino 重构）**：技能包放在 **`skills_dir`**，遵循 **Agent Skills** 目录规范（`SKILL.md` + 可选文件）；**多代理** 下通过 Eino 官方 **`skill`** 工具 **渐进式披露**（按 name 加载）。**`multi_agent.eino_skills`** 控制是否启用、本机文件/Shell 工具、工具名覆盖；**`eino_middleware`** 可选 patch、tool_search、plantask、reduction、断点目录及 Deep 调参。20+ 领域示例仍可绑定角色
|
||||
- 📱 **机器人**：支持钉钉、飞书长连接，在手机端与 CyberStrikeAI 对话（配置与命令详见 [机器人使用说明](docs/robot.md)）
|
||||
- 🐚 **WebShell 管理**：添加与管理 WebShell 连接（兼容冰蝎/蚁剑等），通过虚拟终端执行命令、内置文件管理进行文件操作，并提供按连接维度保存历史的 AI 助手标签页；支持 PHP/ASP/ASPX/JSP 及自定义类型，可配置请求方法与命令参数。
|
||||
|
||||
@@ -226,7 +226,7 @@ go build -o cyberstrike-ai cmd/server/main.go
|
||||
|
||||
### 常用流程
|
||||
- **对话测试**：自然语言触发多步工具编排，SSE 实时输出。
|
||||
- **单代理 / 多代理**：配置 `multi_agent.enabled: true` 后，聊天界面可切换 **单代理**（原有 ReAct 循环）与 **多代理**（Eino DeepAgent + `task` 子代理）。多代理走 `/api/multi-agent/stream`，MCP 工具与单代理同源桥接。
|
||||
- **单代理 / 多代理**：`multi_agent.enabled: true` 后可在聊天中切换 **单代理**（原有 **ReAct**，`/api/agent-loop/stream`）与 **多代理**（`/api/multi-agent/stream`）。多代理在既有 **`deep`**（`task` 子代理）基础上，新增 **`plan_execute`**、**`supervisor`**，由 **`orchestration`** 指定。MCP 工具与单代理同源桥接。
|
||||
- **角色化测试**：从预设的安全测试角色（渗透测试、CTF、Web 应用扫描、API 安全测试等）中选择，自定义 AI 行为和可用工具。每个角色可应用自定义系统提示词，并可限制可用工具列表，实现聚焦的测试场景。
|
||||
- **工具监控**：查看任务队列、执行日志、大文件附件。
|
||||
- **会话历史**：所有对话与工具调用保存在 SQLite，可随时重放。
|
||||
@@ -248,7 +248,7 @@ go build -o cyberstrike-ai cmd/server/main.go
|
||||
- **预设角色**：系统内置 12+ 个预设的安全测试角色（渗透测试、CTF、Web 应用扫描、API 安全测试、二进制分析、云安全审计等），位于 `roles/` 目录。
|
||||
- **自定义提示词**：每个角色可定义 `user_prompt`，会在用户消息前自动添加，引导 AI 采用特定的测试方法和关注重点。
|
||||
- **工具限制**：角色可指定 `tools` 列表，限制可用工具，实现聚焦的测试流程（如 CTF 角色限制为 CTF 专用工具）。
|
||||
- **Skills 集成**：角色可附加安全测试技能。技能 id 会作为提示写入系统提示词；**多代理（DeepAgent）** 会话中由 Eino ADK **`skill`** 工具按需加载包内正文与资源（单代理不含该工具链）。
|
||||
- **Skills 集成**：角色可附加安全测试技能，id 写入提示；**多代理** 下由 Eino **`skill`** 工具 **按需加载**（渐进式披露）。**`multi_agent.eino_skills`** 控制中间件与本机 read_file/glob/grep/write/edit/execute（**Deep / Supervisor** 主/子代理；**plan_execute** 执行器无独立 skill 中间件，见文档）。**单代理 ReAct** 当前不挂载该 Eino skill 链。
|
||||
- **轻松创建角色**：通过在 `roles/` 目录添加 YAML 文件即可创建自定义角色。每个角色定义 `name`、`description`、`user_prompt`、`icon`、`tools`、`skills`、`enabled` 字段。
|
||||
- **Web 界面集成**：在聊天界面通过下拉菜单选择角色。角色选择会影响 AI 行为和可用工具建议。
|
||||
|
||||
@@ -269,20 +269,24 @@ go build -o cyberstrike-ai cmd/server/main.go
|
||||
```
|
||||
2. 重启服务或重新加载配置，角色会出现在角色选择下拉菜单中。
|
||||
|
||||
### 多代理模式（Eino DeepAgent）
|
||||
- **能力说明**：基于 CloudWeGo **Eino** `adk/prebuilt/deep` 的可选路径：**协调主代理**通过内置 **`task`** 工具启动短时**子代理**，各子代理独立推理，工具集来自当前聊天所选角色（与单代理一致来源）。
|
||||
- **Markdown 定义**：在 `agents_dir`（默认 `agents/`，相对 `config.yaml` 所在目录）维护：
|
||||
- **主代理**：固定文件名 `orchestrator.md`，或任意 `.md` 且在 front matter 写 `kind: orchestrator`（**同一目录仅允许一个**主代理）。配置 Deep 的 name/id、description 与可选完整系统提示（正文）；正文为空时依次使用 `multi_agent.orchestrator_instruction`、Eino 内置默认提示。
|
||||
- **子代理**：其余 `*.md`（YAML front matter + 正文作 instruction），不参与主代理定义的文件才会进入 `task` 可选列表。
|
||||
- **界面管理**：**Agents → Agent 管理** 对 Markdown 增删改查；HTTP API 前缀 `/api/multi-agent/markdown-agents`。
|
||||
- **配置项**：`config.yaml` 中 `multi_agent`：`enabled`、`default_mode`（`single` | `multi`）、`robot_use_multi_agent`、`batch_use_multi_agent`、`max_iteration`、`orchestrator_instruction` 等；可选在 YAML 写 `sub_agents` 与目录合并（同 `id` 时以 Markdown 为准）。
|
||||
- **更多细节**：流式事件、机器人与批量任务、排障等见 **[docs/MULTI_AGENT_EINO.md](docs/MULTI_AGENT_EINO.md)**。
|
||||
### 多代理模式（Eino：Deep / Plan-Execute / Supervisor）
|
||||
- **能力说明**：与 **单代理 ReAct** 并存的可选路径，基于 CloudWeGo **Eino** `adk/prebuilt`：**`deep`** — 协调主代理 + **`task`** 子代理；**`plan_execute`** — 规划 / 执行 / 重规划闭环（不使用 YAML/Markdown 子代理列表）；**`supervisor`** — 主代理 **`transfer`** / **`exit`** 调度 Markdown 专家。客户端通过 **`orchestration`** 选 `deep` | `plan_execute` | `supervisor`（缺省 `deep`）。
|
||||
- **Markdown 定义**（`agents_dir`，默认 `agents/`）：
|
||||
- **Deep 主代理**：`orchestrator.md` 或唯一 `kind: orchestrator` 的 `.md`；正文或 `multi_agent.orchestrator_instruction`，再回退 Eino 默认。
|
||||
- **Plan-Execute 主代理**：固定 **`orchestrator-plan-execute.md`**（另可配 `orchestrator_instruction_plan_execute`）。
|
||||
- **Supervisor 主代理**：固定 **`orchestrator-supervisor.md`**（另可配 `orchestrator_instruction_supervisor`）；至少需一名子代理。
|
||||
- **子代理**（**deep** / **supervisor**）：其余 `*.md`；标成 orchestrator 的不会进入 `task` 列表。
|
||||
- **界面管理**：**Agents → Agent 管理**；API `/api/multi-agent/markdown-agents`。
|
||||
- **配置项**：`multi_agent`：`enabled`、`default_mode`、`robot_use_multi_agent`、`batch_use_multi_agent`、`max_iteration`、`plan_execute_loop_max_iterations`、各模式 orchestrator 指令字段、可选 YAML `sub_agents` 与目录合并（同 `id` → Markdown 优先）、**`eino_skills`**、**`eino_middleware`**。
|
||||
- **更多细节**：[docs/MULTI_AGENT_EINO.md](docs/MULTI_AGENT_EINO.md)（流式、机器人、批量、中间件差异）。
|
||||
|
||||
### Skills 技能系统（Agent Skills + Eino）
|
||||
- **目录规范**：与 [Agent Skills](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview) 一致，**仅**需目录下的 **`SKILL.md`**：YAML 头只用官方的 **`name` 与 `description`**，正文为 Markdown。可选同目录其他文件（`FORMS.md`、`REFERENCE.md`、`scripts/*` 等）。**不使用 `SKILL.yaml`**（Claude / Eino 官方均无此文件）；章节、`scripts/` 列表、渐进式行为由运行时从正文与磁盘 **自动推导**。
|
||||
- **Eino**：技能包会被切成 `schema.Document`，供 `FilesystemSkillsRetriever`（`skills.AsEinoRetriever()`）在编排中使用。
|
||||
- **运行侧重构**：**`skills_dir`** 为技能包唯一根目录；**多代理** 通过 Eino 官方 **`skill`** 中间件做 **渐进式披露**（模型按 **name** 调用 `skill`，而非一次性注入全文）。由 **`multi_agent.eino_skills`** 控制：`disable`、`filesystem_tools`（本机读写与 Shell）、`skill_tool_name`。
|
||||
- **Eino / 知识流水线**：技能包可切分为 `schema.Document`，供 `FilesystemSkillsRetriever`（`skills.AsEinoRetriever()`）在 **compose** 图（如索引/编排）中使用。
|
||||
- **提示词**：角色绑定的技能 **id**（文件夹名）会作为推荐写入系统提示；正文默认不整包注入。
|
||||
- **HTTP 管理**：`/api/skills` 列表与 `depth=summary|full`、`section`、`resource_path` 等查询仍用于 Web 与运维；**模型侧**加载技能走多代理内置 **`skill`** 工具，而非 MCP。
|
||||
- **HTTP 管理**：`/api/skills` 列表与 `depth=summary|full`、`section`、`resource_path` 等仍用于 Web 与运维；**模型侧** 多代理走 **`skill`** 工具，而非 MCP。
|
||||
- **可选 `eino_middleware`**：如 `tool_search`（动态工具列表）、`patch_tool_calls`、`plantask`（结构化任务；默认落在 `skills_dir` 下子目录）、`reduction`、`checkpoint_dir`、Deep 输出键 / 模型重试 / task 描述前缀等，见 `config.yaml` 与 `internal/config/config.go`。
|
||||
- **自带示例**：`skills/cyberstrike-eino-demo/`；说明见 `skills/README.md`。
|
||||
|
||||
**新建技能：**
|
||||
@@ -520,7 +524,10 @@ multi_agent:
|
||||
default_mode: "single" # single | multi（开启多代理时的界面默认模式）
|
||||
robot_use_multi_agent: false
|
||||
batch_use_multi_agent: false
|
||||
orchestrator_instruction: "" # 可选；orchestrator.md 正文为空时使用
|
||||
orchestrator_instruction: "" # Deep；orchestrator.md 正文为空时使用
|
||||
# orchestrator_instruction_plan_execute / orchestrator_instruction_supervisor 可选
|
||||
# eino_skills: { disable: false, filesystem_tools: true, skill_tool_name: skill }
|
||||
# eino_middleware: 可选 patch_tool_calls、tool_search、plantask、reduction、checkpoint_dir 等
|
||||
```
|
||||
|
||||
### 工具模版示例（`tools/nmap.yaml`）
|
||||
@@ -565,7 +572,7 @@ enabled: true
|
||||
|
||||
## 相关文档
|
||||
|
||||
- [多代理模式（Eino）](docs/MULTI_AGENT_EINO.md)：DeepAgent 编排、`agents/*.md`、接口与流式说明。
|
||||
- [多代理模式（Eino）](docs/MULTI_AGENT_EINO.md)：**Deep**、**Plan-Execute**、**Supervisor**、`agents/*.md`、`eino_skills` / `eino_middleware`、接口与流式说明。
|
||||
- [机器人使用说明（钉钉 / 飞书）](docs/robot.md)：在手机端通过钉钉、飞书与 CyberStrikeAI 对话的完整配置步骤、命令与排查说明，**建议按该文档操作以避免走弯路**。
|
||||
|
||||
## 项目结构
|
||||
|
||||
Reference in New Issue
Block a user