Merge branch 'develop' into prompt_template

Author: 2h0u4n

.gitignore

@@ -21,6 +21,7 @@ docker/volumes/db/data
 docker/.env
 docker/.run
 docker/deploy.options
+k8s/helm/.deploy.options
 
 frontend_standalone/
 .pnpm-store/
@@ -32,4 +33,4 @@ model-assets/
 *coverage_html
 *.pytest_cache
 *.coverage
-*coverage.xml
+*coverage.xml

LICENSE

@@ -1,20 +1,21 @@
-# Nexent Open Source License
-
-Nexent is licensed under the MIT License, with the following additional conditions:
-
-Nexent is permitted to be used commercially, including as a backend service for other applications or as an application development platform for enterprises. However, when the following conditions are met, you must contact the producer to obtain a commercial license:
-
-a. Multi-tenant SaaS service: Unless explicitly authorized by Nexent in writing, you may not use the Nexent source code to operate a multi-tenant SaaS service.
-b. LOGO and copyright information: In the process of using Nexent's frontend, you may not remove or modify the LOGO or copyright information in the Nexent console or applications. This restriction is inapplicable to uses of Nexent that do not involve its frontend.
-
-Please contact zhenggaoqi@huawei.com by email to inquire about licensing matters.
-
-As a contributor, you should agree that:
-
-a. The producer can adjust the open-source agreement to be more strict or relaxed as deemed necessary.
-b. Your contributed code may be used for commercial purposes, such as Nexent's cloud business.
-
-Apart from the specific conditions mentioned above, all other rights and restrictions follow the MIT License.
-Detailed information about the MIT License can be found at: https://opensource.org/licenses/MIT
-
-Copyright © 2025 Huawei Technologies Co., Ltd.
+MIT License
+
+Copyright (c) 2025 Huawei Technologies Co., Ltd. All rights reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -7,7 +7,7 @@
 [![Docker Pulls](https://img.shields.io/docker/pulls/nexent/nexent?logo=docker&label=DockerPull)](https://hub.docker.com/repositories/nexent)
 [![Codecov (with branch)](https://img.shields.io/codecov/c/github/ModelEngine-Group/nexent/develop?logo=codecov&color=green)](https://codecov.io/gh/ModelEngine-Group/nexent)
 
-Nexent is a zero-code platform for auto-generating agents — no orchestration, no complex drag-and-drop required, using pure language to develop any agent you want. Built on the MCP ecosystem with rich tool integration, Nexent also provides various built-in agents to meet your intelligent service needs in different scenarios such as work, travel, and daily life. Nexent offers powerful capabilities for agent running control, multi-agent collaboration, data processing and knowledge tracing, multimodal dialogue, and batch scaling.
+Nexent is a zero-code platform for auto-generating production-grade AI agents, built on **Harness Engineering** principles. It provides unified tools, skills, memory, and orchestration with built-in constraints, feedback loops, and control planes — no orchestration, no complex drag-and-drop required, using pure language to develop any agent you want.
 
 > One prompt. Endless reach.
 
@@ -58,7 +58,7 @@ Most of all, we need visibility. Star ⭐ and watch the repo, share it with frie
 
 ## 💬 Community & contact
 
-- Browse the [Documentation](https://modelengine-group.github.io/nexent) for more information.  
+- Browse the [Documentation](https://modelengine-group.github.io/nexent) for more information.
 - Join our [Discord community](https://discord.gg/tb5H3S3wyv) to chat with other developers and get help!
 - Conntact us by Wechat, find our QR Code in our [website](https://nexent.tech/en/contact)
 
@@ -119,5 +119,4 @@ Prefer to run Nexent from source code? Follow our [Developer Guide](https://mode
 
 # 📄 License
 
-Nexent is licensed under the [MIT](LICENSE) with additional conditions. Please read the [LICENSE](LICENSE) file for details.
-
+Nexent is licensed under the [MIT License](LICENSE).

README_CN.md

@@ -7,7 +7,7 @@
 [![Docker Pulls](https://img.shields.io/docker/pulls/nexent/nexent?logo=docker&label=DockerPull)](https://hub.docker.com/repositories/nexent)
 [![Codecov (with branch)](https://img.shields.io/codecov/c/github/ModelEngine-Group/nexent/develop?logo=codecov&color=green)](https://codecov.io/gh/ModelEngine-Group/nexent)
 
-Nexent 是一个零代码智能体自动生成平台 —— 无需编排，无需复杂的拖拉拽操作，使用纯语言开发你想要的任何智能体。基于MCP生态，具备丰富的工具集成，同时提供多种自带智能体，满足你的工作、旅行、生活等不同场景的智能服务需要。Nexent 还提供强大的智能体运行控制、多智能体协作、数据处理和知识溯源、多模态对话、批量扩展能力。
+Nexent 是一个基于 **Harness Engineering** 原则打造的零代码智能体自动生成平台。集统一工具、技能、记忆和编排能力于一体，内置约束机制、反馈循环和控制平面。无需编排，无需复杂的拖拉拽操作，使用纯语言开发你想要的任何智能体。
 
 > 一个提示词，无限种可能。
 
@@ -58,7 +58,7 @@ bash deploy.sh
 
 ## 💬 社区与联系方式
 
-- 浏览 [文档](https://modelengine-group.github.io/nexent) 了解更多信息。  
+- 浏览 [文档](https://modelengine-group.github.io/nexent) 了解更多信息。
 - 加入我们的 [Discord 社区](https://discord.gg/tb5H3S3wyv) 与其他开发者交流并获取帮助！
 - 通过微信联系我们，在我们的[网站](https://nexent.tech/zh/contact)找到二维码
 
@@ -119,4 +119,4 @@ bash deploy.sh
 
 # 📄 许可证
 
-Nexent 采用 [MIT](LICENSE) 许可证，并附有额外条件。请阅读 [LICENSE](LICENSE) 文件了解详情。
+Nexent 采用 [MIT 许可证](LICENSE)。

backend/agents/create_agent_info.py

@@ -1,10 +1,10 @@
 import threading
 import logging
+from typing import List
 from urllib.parse import urljoin
 from datetime import datetime
 
 from jinja2 import Template, StrictUndefined
-from smolagents.utils import BASE_BUILTIN_MODULES
 from nexent.core.utils.observer import MessageObserver
 from nexent.core.agents.agent_model import AgentRunInfo, ModelConfig, AgentConfig, ToolConfig
 from nexent.memory.memory_service import search_memory_in_levels
@@ -14,6 +14,7 @@
     ElasticSearchService,
     get_vector_db_core,
     get_embedding_model,
+    get_rerank_model,
 )
 from services.remote_mcp_service import get_remote_mcp_server_list
 from services.memory_config_service import build_memory_context
@@ -27,11 +28,119 @@
 from utils.prompt_template_utils import get_agent_prompt_template
 from utils.config_utils import tenant_config_manager, get_model_name_from_config
 from consts.const import LOCAL_MCP_SERVER, MODEL_CONFIG_MAPPING, LANGUAGE, DATA_PROCESS_SERVICE
+import re
 
 logger = logging.getLogger("create_agent_info")
 logger.setLevel(logging.DEBUG)
 
 
+def _get_skills_for_template(
+    agent_id: int,
+    tenant_id: str,
+    version_no: int = 0
+) -> List[dict]:
+    """Get skills list for prompt template injection.
+
+    Args:
+        agent_id: Agent ID
+        tenant_id: Tenant ID
+        version_no: Version number
+
+    Returns:
+        List of skill dicts with name and description
+    """
+    try:
+        from services.skill_service import SkillService
+        skill_service = SkillService()
+        enabled_skills = skill_service.get_enabled_skills_for_agent(
+            agent_id=agent_id,
+            tenant_id=tenant_id,
+            version_no=version_no
+        )
+        return [
+            {"name": s.get("name", ""), "description": s.get("description", "")}
+            for s in enabled_skills
+        ]
+    except Exception as e:
+        logger.warning(f"Failed to get skills for template: {e}")
+        return []
+
+
+def _get_skill_script_tools(
+    agent_id: int,
+    tenant_id: str,
+    version_no: int = 0
+) -> List[ToolConfig]:
+    """Get tool config for skill script execution and skill reading.
+
+    Args:
+        agent_id: Agent ID for filtering available skills in error messages.
+        tenant_id: Tenant ID for filtering available skills in error messages.
+        version_no: Version number for filtering available skills.
+
+    Returns:
+        List of ToolConfig for skill execution and reading tools
+    """
+    from consts.const import CONTAINER_SKILLS_PATH
+
+    skill_context = {
+        "agent_id": agent_id,
+        "tenant_id": tenant_id,
+        "version_no": version_no,
+    }
+
+    try:
+        return [
+            ToolConfig(
+                class_name="RunSkillScriptTool",
+                name="run_skill_script",
+                description="Execute a skill script with given parameters. Use this to run Python or shell scripts that are part of a skill.",
+                inputs='{"skill_name": "str", "script_path": "str", "params": "dict"}',
+                output_type="string",
+                params={"local_skills_dir": CONTAINER_SKILLS_PATH},
+                source="builtin",
+                usage="builtin",
+                metadata=skill_context,
+            ),
+            ToolConfig(
+                class_name="ReadSkillMdTool",
+                name="read_skill_md",
+                description="Read skill execution guide and optional additional files. Always reads SKILL.md first, then optionally reads additional files.",
+                inputs='{"skill_name": "str", "additional_files": "list[str]"}',
+                output_type="string",
+                params={"local_skills_dir": CONTAINER_SKILLS_PATH},
+                source="builtin",
+                usage="builtin",
+                metadata=skill_context,
+            ),
+            ToolConfig(
+                class_name="ReadSkillConfigTool",
+                name="read_skill_config",
+                description="Read the config.yaml file from a skill directory. Returns JSON containing configuration variables needed for skill workflows.",
+                inputs='{"skill_name": "str"}',
+                output_type="string",
+                params={"local_skills_dir": CONTAINER_SKILLS_PATH},
+                source="builtin",
+                usage="builtin",
+                metadata=skill_context,
+            ),
+            ToolConfig(
+                class_name="WriteSkillFileTool",
+                name="write_skill_file",
+                description="Write content to a file within a skill directory. Creates parent directories if they do not exist.",
+                inputs='{"skill_name": "str", "file_path": "str", "content": "str"}',
+                output_type="string",
+                params={"local_skills_dir": CONTAINER_SKILLS_PATH},
+                source="builtin",
+                usage="builtin",
+                metadata=skill_context,
+            )
+        ]
+    except Exception as e:
+        logger.warning(f"Failed to load skill script tool: {e}")
+        return []
+
+
 async def create_model_config_list(tenant_id):
     records = get_model_records({"model_type": "llm"}, tenant_id)
     model_list = []
@@ -171,22 +280,24 @@ async def create_agent_config(
         logger.error(f"Failed to build knowledge base summary: {e}")
 
     # Assemble system_prompt
-    if duty_prompt or constraint_prompt or few_shots_prompt:
-        system_prompt = Template(prompt_template["system_prompt"], undefined=StrictUndefined).render({
-            "duty": duty_prompt,
-            "constraint": constraint_prompt,
-            "few_shots": few_shots_prompt,
-            "tools": {tool.name: tool for tool in tool_list},
-            "managed_agents": {agent.name: agent for agent in managed_agents},
-            "authorized_imports": str(BASE_BUILTIN_MODULES),
-            "APP_NAME": app_name,
-            "APP_DESCRIPTION": app_description,
-            "memory_list": memory_list,
-            "knowledge_base_summary": knowledge_base_summary,
-            "time": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
-        })
-    else:
-        system_prompt = agent_info.get("prompt", "")
+    # Get skills list for prompt template
+    skills = _get_skills_for_template(agent_id, tenant_id, version_no)
+
+    render_kwargs = {
+        "duty": duty_prompt,
+        "constraint": constraint_prompt,
+        "few_shots": few_shots_prompt,
+        "tools": {tool.name: tool for tool in tool_list},
+        "skills": skills,
+        "managed_agents": {agent.name: agent for agent in managed_agents},
+        "APP_NAME": app_name,
+        "APP_DESCRIPTION": app_description,
+        "memory_list": memory_list,
+        "knowledge_base_summary": knowledge_base_summary,
+        "time": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
+        "user_id": user_id,
+    }
+    system_prompt = Template(prompt_template["system_prompt"], undefined=StrictUndefined).render(render_kwargs)
 
     model_id_to_use = override_model_id if override_model_id else agent_info.get("model_id")
     if model_id_to_use is not None:
@@ -200,9 +311,10 @@ async def create_agent_config(
         prompt_templates=await prepare_prompt_templates(
             is_manager=len(managed_agents) > 0,
             system_prompt=system_prompt,
-            language=language
+            language=language,
+            agent_id=agent_id
         ),
-        tools=tool_list,
+        tools=tool_list + _get_skill_script_tools(agent_id, tenant_id, version_no),
         max_steps=agent_info.get("max_steps", 10),
         model_name=model_name,
         provide_run_summary=agent_info.get("provide_run_summary", False),
@@ -240,11 +352,32 @@ async def create_tool_config_list(agent_id, tenant_id, user_id, version_no: int
                     tool_config.metadata = langchain_tool
                     break
 
-        # special logic for knowledge base search tool
+        # special logic for search tools that may use reranking models
         if tool_config.class_name == "KnowledgeBaseSearchTool":
-           tool_config.metadata = {
+            rerank = param_dict.get("rerank", False)
+            rerank_model_name = param_dict.get("rerank_model_name", "")
+            rerank_model = None
+            if rerank and rerank_model_name:
+                rerank_model = get_rerank_model(
+                    tenant_id=tenant_id, model_name=rerank_model_name
+                )
+
+            tool_config.metadata = {
                 "vdb_core": get_vector_db_core(),
                 "embedding_model": get_embedding_model(tenant_id=tenant_id),
+                "rerank_model": rerank_model,
+            }
+        elif tool_config.class_name in ["DifySearchTool", "DataMateSearchTool"]:
+            rerank = param_dict.get("rerank", False)
+            rerank_model_name = param_dict.get("rerank_model_name", "")
+            rerank_model = None
+            if rerank and rerank_model_name:
+                rerank_model = get_rerank_model(
+                    tenant_id=tenant_id, model_name=rerank_model_name
+                )
+
+            tool_config.metadata = {
+                "rerank_model": rerank_model,
             }
         elif tool_config.class_name == "AnalyzeTextFileTool":
             tool_config.metadata = {
@@ -299,20 +432,27 @@ async def discover_langchain_tools():
     return langchain_tools
 
 
-async def prepare_prompt_templates(is_manager: bool, system_prompt: str, language: str = 'zh'):
+async def prepare_prompt_templates(
+    is_manager: bool,
+    system_prompt: str,
+    language: str = 'zh',
+    agent_id: int = None,
+):
     """
     Prepare prompt templates, support multiple languages
 
     Args:
         is_manager: Whether it is a manager mode
         system_prompt: System prompt content
         language: Language code ('zh' or 'en')
+        agent_id: Agent ID for fetching skill instances
 
     Returns:
         dict: Prompt template configuration
     """
     prompt_templates = get_agent_prompt_template(is_manager, language)
     prompt_templates["system_prompt"] = system_prompt
+
     return prompt_templates