Feat skills

README.md

@@ -75,7 +75,7 @@ Built on [CAMEL-AI][camel-site]'s acclaimed open-source project, our system intr
 - [📄 Open Source License](#-open-source-license)
 - [🌐 Community & contact](#-community--contact)
 
-#### 
+####
 
 <br/>
 

README_CN.md

@@ -74,7 +74,7 @@
 - [📄 开源许可证](#-%E5%BC%80%E6%BA%90%E8%AE%B8%E5%8F%AF%E8%AF%81)
 - [🌐 社区与联系](#-%E7%A4%BE%E5%8C%BA%E4%B8%8E%E8%81%94%E7%B3%BB)
 
-#### 
+####
 
 <br/>
 

README_JA.md

@@ -73,7 +73,7 @@
 - [📄 オープンソースライセンス](#-%E3%82%AA%E3%83%BC%E3%83%97%E3%83%B3%E3%82%BD%E3%83%BC%E3%82%B9%E3%83%A9%E3%82%A4%E3%82%BB%E3%83%B3%E3%82%B9)
 - [🌐 コミュニティ & お問い合わせ](#-%E3%82%B3%E3%83%9F%E3%83%A5%E3%83%8B%E3%83%86%E3%82%A3--%E3%81%8A%E5%95%8F%E3%81%84%E5%90%88%E3%82%8F%E3%81%9B)
 
-#### 
+####
 
 <br/>
 

README_PT-BR.md

@@ -74,7 +74,7 @@ Construído sobre o aclamado projeto open source da [CAMEL-AI][camel-site], noss
 - [📄 Licença Open Source](#-licen%C3%A7a-open-source)
 - [🌐 Comunidade & Contato](#-comunidade--contato)
 
-#### 
+####
 
 <br/>
 

backend/app/agent/factory/browser.py

@@ -27,6 +27,7 @@
 # TODO: Remove NoteTakingToolkit and use TerminalToolkit instead
 from app.agent.toolkit.note_taking_toolkit import NoteTakingToolkit
 from app.agent.toolkit.search_toolkit import SearchToolkit
+from app.agent.toolkit.skill_toolkit import SkillToolkit
 from app.agent.toolkit.terminal_toolkit import TerminalToolkit
 from app.agent.utils import NOW_STR
 from app.component.environment import env
@@ -97,6 +98,13 @@ def browser_agent(options: Chat):
     )
     note_toolkit = message_integration.register_toolkits(note_toolkit)
 
+    skill_toolkit = SkillToolkit(
+        options.project_id,
+        Agents.browser_agent,
+        working_directory=working_directory,
+    )
+    skill_toolkit = message_integration.register_toolkits(skill_toolkit)
+
     search_tools = SearchToolkit.get_can_use_tools(options.project_id)
     if search_tools:
         search_tools = message_integration.register_functions(search_tools)
@@ -111,6 +119,7 @@ def browser_agent(options: Chat):
         *terminal_toolkit,
         *note_toolkit.get_tools(),
         *search_tools,
+        *skill_toolkit.get_tools(),
     ]
 
     system_message = BROWSER_SYS_PROMPT.format(
@@ -135,6 +144,7 @@ def browser_agent(options: Chat):
             HumanToolkit.toolkit_name(),
             NoteTakingToolkit.toolkit_name(),
             TerminalToolkit.toolkit_name(),
+            SkillToolkit.toolkit_name(),
         ],
         toolkits_to_register_agent=[web_toolkit_for_agent_registration],
         enable_snapshot_clean=True,

backend/app/agent/factory/developer.py

@@ -25,6 +25,7 @@
 # TODO: Remove NoteTakingToolkit and use TerminalToolkit instead
 from app.agent.toolkit.note_taking_toolkit import NoteTakingToolkit
 from app.agent.toolkit.screenshot_toolkit import ScreenshotToolkit
+from app.agent.toolkit.skill_toolkit import SkillToolkit
 from app.agent.toolkit.terminal_toolkit import TerminalToolkit
 from app.agent.toolkit.web_deploy_toolkit import WebDeployToolkit
 from app.agent.utils import NOW_STR
@@ -70,6 +71,13 @@ async def developer_agent(options: Chat):
     )
     terminal_toolkit = message_integration.register_toolkits(terminal_toolkit)
 
+    skill_toolkit = SkillToolkit(
+        options.project_id,
+        Agents.developer_agent,
+        working_directory=working_directory,
+    )
+    skill_toolkit = message_integration.register_toolkits(skill_toolkit)
+
     tools = [
         *HumanToolkit.get_can_use_tools(
             options.project_id, Agents.developer_agent
@@ -78,6 +86,7 @@ async def developer_agent(options: Chat):
         *web_deploy_toolkit.get_tools(),
         *terminal_toolkit.get_tools(),
         *screenshot_toolkit.get_tools(),
+        *skill_toolkit.get_tools(),
     ]
     system_message = DEVELOPER_SYS_PROMPT.format(
         platform_system=platform.system(),
@@ -99,5 +108,6 @@ async def developer_agent(options: Chat):
             TerminalToolkit.toolkit_name(),
             NoteTakingToolkit.toolkit_name(),
             WebDeployToolkit.toolkit_name(),
+            SkillToolkit.toolkit_name(),
         ],
     )

backend/app/agent/factory/document.py

@@ -28,6 +28,7 @@
 # TODO: Remove NoteTakingToolkit and use TerminalToolkit instead
 from app.agent.toolkit.note_taking_toolkit import NoteTakingToolkit
 from app.agent.toolkit.pptx_toolkit import PPTXToolkit
+from app.agent.toolkit.skill_toolkit import SkillToolkit
 from app.agent.toolkit.terminal_toolkit import TerminalToolkit
 from app.agent.utils import NOW_STR
 from app.model.chat import Chat
@@ -82,6 +83,13 @@ async def document_agent(options: Chat):
         options.project_id, options.get_bun_env()
     )
 
+    skill_toolkit = SkillToolkit(
+        options.project_id,
+        Agents.document_agent,
+        working_directory=working_directory,
+    )
+    skill_toolkit = message_integration.register_toolkits(skill_toolkit)
+
     tools = [
         *file_write_toolkit.get_tools(),
         *pptx_toolkit.get_tools(),
@@ -93,6 +101,7 @@ async def document_agent(options: Chat):
         *note_toolkit.get_tools(),
         *terminal_toolkit.get_tools(),
         *google_drive_tools,
+        *skill_toolkit.get_tools(),
     ]
     system_message = DOCUMENT_SYS_PROMPT.format(
         platform_system=platform.system(),
@@ -118,5 +127,6 @@ async def document_agent(options: Chat):
             NoteTakingToolkit.toolkit_name(),
             TerminalToolkit.toolkit_name(),
             GoogleDriveMCPToolkit.toolkit_name(),
+            SkillToolkit.toolkit_name(),
         ],
     )

backend/app/agent/factory/multi_modal.py

@@ -29,6 +29,7 @@
 from app.agent.toolkit.note_taking_toolkit import NoteTakingToolkit
 from app.agent.toolkit.openai_image_toolkit import OpenAIImageToolkit
 from app.agent.toolkit.search_toolkit import SearchToolkit
+from app.agent.toolkit.skill_toolkit import SkillToolkit
 from app.agent.toolkit.terminal_toolkit import TerminalToolkit
 from app.agent.toolkit.video_download_toolkit import VideoDownloaderToolkit
 from app.agent.utils import NOW_STR
@@ -75,6 +76,13 @@ def multi_modal_agent(options: Chat):
         working_directory=working_directory,
     )
     note_toolkit = message_integration.register_toolkits(note_toolkit)
+
+    skill_toolkit = SkillToolkit(
+        options.project_id,
+        Agents.multi_modal_agent,
+        working_directory=working_directory,
+    )
+    skill_toolkit = message_integration.register_toolkits(skill_toolkit)
     tools = [
         *video_download_toolkit.get_tools(),
         *image_analysis_toolkit.get_tools(),
@@ -83,6 +91,7 @@ def multi_modal_agent(options: Chat):
         ),
         *terminal_toolkit.get_tools(),
         *note_toolkit.get_tools(),
+        *skill_toolkit.get_tools(),
     ]
     if options.is_cloud():
         # TODO: check llm has this model
@@ -147,5 +156,6 @@ def multi_modal_agent(options: Chat):
             TerminalToolkit.toolkit_name(),
             NoteTakingToolkit.toolkit_name(),
             SearchToolkit.toolkit_name(),
+            SkillToolkit.toolkit_name(),
         ],
     )

backend/app/agent/prompt.py

@@ -142,6 +142,13 @@
 
 <capabilities>
 Your capabilities include:
+- **Skills System**: You have access to a library of specialized skills that
+  provide expert guidance for specific tasks. When a skill is referenced with
+  double curly braces (e.g., {{pdf}} or {{data-analyzer}}), you should use
+  `list_skills` to discover available skills and `load_skill` to retrieve the
+  skill's full content. Skills contain tested code examples, best practices,
+  and detailed instructions that you MUST follow as your PRIMARY reference when
+  working on tasks that mention them.
 - Video & Audio Analysis:
     - Download videos from URLs for analysis.
     - Transcribe speech from audio files to text with high accuracy
@@ -263,6 +270,13 @@
 
 <capabilities>
 Your capabilities include:
+- **Skills System**: You have access to a library of specialized skills that
+  provide expert guidance for specific tasks. When a skill is referenced with
+  double curly braces (e.g., {{pdf}} or {{data-analyzer}}), you should use
+  `list_skills` to discover available skills and `load_skill` to retrieve the
+  skill's full content. Skills contain tested code examples, best practices,
+  and detailed instructions that you MUST follow as your PRIMARY reference when
+  working on tasks that mention them.
 - Document Reading:
     - Read and understand the content of various file formats including
         - PDF (.pdf)
@@ -413,6 +427,13 @@
 
 <capabilities>
 Your capabilities are extensive and powerful:
+- **Skills System**: You have access to a library of specialized skills that
+  provide expert guidance for specific tasks. When a skill is referenced with
+  double curly braces (e.g., {{pdf}} or {{data-analyzer}}), you should use
+  `list_skills` to discover available skills and `load_skill` to retrieve the
+  skill's full content. Skills contain tested code examples, best practices,
+  and detailed instructions that you MUST follow as your PRIMARY reference when
+  working on tasks that mention them.
 - **Unrestricted Code Execution**: You can write and execute code in any
   language to solve a task. You MUST first save your code to a file (e.g.,
   `script.py`) and then run it from the terminal (e.g.,
@@ -581,6 +602,13 @@
 
 <capabilities>
 Your capabilities include:
+- **Skills System**: You have access to a library of specialized skills that
+  provide expert guidance for specific tasks. When a skill is referenced with
+  double curly braces (e.g., {{pdf}} or {{data-analyzer}}), you should use
+  `list_skills` to discover available skills and `load_skill` to retrieve the
+  skill's full content. Skills contain tested code examples, best practices,
+  and detailed instructions that you MUST follow as your PRIMARY reference when
+  working on tasks that mention them.
 - Search and get information from the web using the search tools.
 - Use the rich browser related toolset to investigate websites.
 - Use the terminal tools to perform local operations. **IMPORTANT:** Before the