agno 2.3.21__tar.gz → 2.3.23__tar.gz

{agno-2.3.21 → agno-2.3.23}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agno
-Version: 2.3.21
+Version: 2.3.23
 Summary: Agno: a lightweight library for building Multi-Agent Systems
 Author-email: Ashpreet Bedi <ashpreet@agno.com>
 Project-URL: homepage, https://agno.com
@@ -157,7 +157,7 @@ Requires-Dist: mcp>=1.9.2; extra == "mcp"
 Provides-Extra: mem0
 Requires-Dist: mem0ai; extra == "mem0"
 Provides-Extra: memori
-Requires-Dist: memorisdk==3.0.5; extra == "memori"
+Requires-Dist: memori>=3.0.5; extra == "memori"
 Provides-Extra: newspaper
 Requires-Dist: newspaper4k; extra == "newspaper"
 Requires-Dist: lxml_html_clean; extra == "newspaper"
@@ -425,23 +425,22 @@ Dynamic: license-file
 
 Agno is a multi-agent framework, runtime, and control plane. Use it to build private and secure AI products that run in your cloud.
 
-- **Build** agents, teams, and workflows with memory, knowledge, guardrails, and 100+ toolkits.
+- **Build** agents, teams, and workflows with memory, knowledge, guardrails and 100+ integrations.
 - **Run** in production with a stateless FastAPI runtime. Horizontally scalable.
 - **Manage** with a control plane that connects directly to your runtime — no data leaves your environment.
 
 ## Why Agno?
 
-- **Your cloud, your data:** AgentOS runs entirely in your infrastructure. Zero data leaves your environment.
-- **Production-ready from day one:** Pre-built FastAPI runtime with SSE endpoints, ready to deploy.
-- **Actually fast:** 529× faster than LangGraph, 24× lower memory. Matters at scale.
+- **Your cloud, your data:** Runs entirely in your infrastructure. Nothing leaves your environment.
+- **Ready for production on day one:** Pre-built FastAPI runtime with SSE endpoints, ready to deploy.
+- **Incredibly fast:** 529× faster than LangGraph, 24× lower memory.
 
 ## Getting Started
 
-New to Agno? Start with the [getting started guide](https://github.com/agno-agi/agno/tree/main/cookbook/00_getting_started).
+Start with the [getting started guide](https://github.com/agno-agi/agno/tree/main/cookbook/00_getting_started), then:
 
-Then:
 - Browse the [cookbooks](https://github.com/agno-agi/agno/tree/main/cookbook) for real-world examples
-- Read the [docs](https://docs.agno.com) to learn more.
+- Read the [docs](https://docs.agno.com) to learn more
 
 ## Resources
 
@@ -509,26 +508,26 @@ This isn't a privacy mode or enterprise add-on. It's how Agno works.
 
 ## Features
 
-**Core**
+### Core:
 - Model agnostic — works with OpenAI, Anthropic, Google, local models, whatever
 - Type-safe I/O with `input_schema` and `output_schema`
 - Async-first, built for long-running tasks
 - Natively multimodal (text, images, audio, video, files)
 
-**Memory & Knowledge**
+### Memory & Knowledge:
 - Persistent storage for session history and state
 - User memory that persists across sessions
 - Agentic RAG with 20+ vector stores, hybrid search, reranking
 - Culture — shared long-term memory across agents
 
-**Execution**
+### Execution:
 - Human-in-the-loop (confirmations, approvals, overrides)
 - Guardrails for validation and security
 - Pre/post hooks for the agent lifecycle
 - First-class MCP and A2A support
 - 100+ built-in toolkits
 
-**Production**
+### Production:
 - Ready-to-use FastAPI runtime
 - Integrated control plane UI
 - Evals for accuracy, performance, latency
@@ -546,7 +545,7 @@ We're obsessive about performance because agent workloads spawn hundreds of inst
 | Instantiation | **3μs** | 1,587μs (529× slower) | 170μs (57× slower) | 210μs (70× slower) |
 | Memory | **6.6 KiB** | 161 KiB (24× higher) | 29 KiB (4× higher) | 66 KiB (10× higher) |
 
-Run the benchmarks yourself: [`cookbook/evals/performance`](https://github.com/agno-agi/agno/tree/main/cookbook/evals/performance)
+Run the benchmarks yourself: [`cookbook/12_evals/performance`](https://github.com/agno-agi/agno/tree/main/cookbook/12_evals/performance)
 
 https://github.com/user-attachments/assets/54b98576-1859-4880-9f2d-15e1a426719d
 

{agno-2.3.21 → agno-2.3.23}/README.md

@@ -21,23 +21,22 @@
 
 Agno is a multi-agent framework, runtime, and control plane. Use it to build private and secure AI products that run in your cloud.
 
-- **Build** agents, teams, and workflows with memory, knowledge, guardrails, and 100+ toolkits.
+- **Build** agents, teams, and workflows with memory, knowledge, guardrails and 100+ integrations.
 - **Run** in production with a stateless FastAPI runtime. Horizontally scalable.
 - **Manage** with a control plane that connects directly to your runtime — no data leaves your environment.
 
 ## Why Agno?
 
-- **Your cloud, your data:** AgentOS runs entirely in your infrastructure. Zero data leaves your environment.
-- **Production-ready from day one:** Pre-built FastAPI runtime with SSE endpoints, ready to deploy.
-- **Actually fast:** 529× faster than LangGraph, 24× lower memory. Matters at scale.
+- **Your cloud, your data:** Runs entirely in your infrastructure. Nothing leaves your environment.
+- **Ready for production on day one:** Pre-built FastAPI runtime with SSE endpoints, ready to deploy.
+- **Incredibly fast:** 529× faster than LangGraph, 24× lower memory.
 
 ## Getting Started
 
-New to Agno? Start with the [getting started guide](https://github.com/agno-agi/agno/tree/main/cookbook/00_getting_started).
+Start with the [getting started guide](https://github.com/agno-agi/agno/tree/main/cookbook/00_getting_started), then:
 
-Then:
 - Browse the [cookbooks](https://github.com/agno-agi/agno/tree/main/cookbook) for real-world examples
-- Read the [docs](https://docs.agno.com) to learn more.
+- Read the [docs](https://docs.agno.com) to learn more
 
 ## Resources
 
@@ -105,26 +104,26 @@ This isn't a privacy mode or enterprise add-on. It's how Agno works.
 
 ## Features
 
-**Core**
+### Core:
 - Model agnostic — works with OpenAI, Anthropic, Google, local models, whatever
 - Type-safe I/O with `input_schema` and `output_schema`
 - Async-first, built for long-running tasks
 - Natively multimodal (text, images, audio, video, files)
 
-**Memory & Knowledge**
+### Memory & Knowledge:
 - Persistent storage for session history and state
 - User memory that persists across sessions
 - Agentic RAG with 20+ vector stores, hybrid search, reranking
 - Culture — shared long-term memory across agents
 
-**Execution**
+### Execution:
 - Human-in-the-loop (confirmations, approvals, overrides)
 - Guardrails for validation and security
 - Pre/post hooks for the agent lifecycle
 - First-class MCP and A2A support
 - 100+ built-in toolkits
 
-**Production**
+### Production:
 - Ready-to-use FastAPI runtime
 - Integrated control plane UI
 - Evals for accuracy, performance, latency
@@ -142,7 +141,7 @@ We're obsessive about performance because agent workloads spawn hundreds of inst
 | Instantiation | **3μs** | 1,587μs (529× slower) | 170μs (57× slower) | 210μs (70× slower) |
 | Memory | **6.6 KiB** | 161 KiB (24× higher) | 29 KiB (4× higher) | 66 KiB (10× higher) |
 
-Run the benchmarks yourself: [`cookbook/evals/performance`](https://github.com/agno-agi/agno/tree/main/cookbook/evals/performance)
+Run the benchmarks yourself: [`cookbook/12_evals/performance`](https://github.com/agno-agi/agno/tree/main/cookbook/12_evals/performance)
 
 https://github.com/user-attachments/assets/54b98576-1859-4880-9f2d-15e1a426719d
 

{agno-2.3.21 → agno-2.3.23}/agno/agent/agent.py

@@ -61,6 +61,9 @@ from agno.run.agent import (
     RunOutput,
     RunOutputEvent,
 )
+from agno.run.cancel import (
+    acancel_run as acancel_run_global,
+)
 from agno.run.cancel import (
     acleanup_run,
     araise_if_cancelled,
@@ -77,6 +80,7 @@ from agno.run.requirement import RunRequirement
 from agno.run.team import TeamRunOutputEvent
 from agno.session import AgentSession, SessionSummaryManager, TeamSession, WorkflowSession
 from agno.session.summary import SessionSummary
+from agno.skills import Skills
 from agno.tools import Toolkit
 from agno.tools.function import Function
 from agno.utils.agent import (
@@ -264,6 +268,10 @@ class Agent:
     knowledge_retriever: Optional[Callable[..., Optional[List[Union[Dict, str]]]]] = None
     references_format: Literal["json", "yaml"] = "json"
 
+    # --- Skills ---
+    # Skills provide structured instructions, reference docs, and scripts for agents
+    skills: Optional[Skills] = None
+
     # --- Agent Tools ---
     # A list of tools provided to the Model.
     # Tools are functions the model may generate JSON inputs for.
@@ -487,6 +495,7 @@ class Agent:
         add_knowledge_to_context: bool = False,
         knowledge_retriever: Optional[Callable[..., Optional[List[Union[Dict, str]]]]] = None,
         references_format: Literal["json", "yaml"] = "json",
+        skills: Optional[Skills] = None,
         metadata: Optional[Dict[str, Any]] = None,
         tools: Optional[Sequence[Union[Toolkit, Callable, Function, Dict]]] = None,
         tool_call_limit: Optional[int] = None,
@@ -610,6 +619,8 @@ class Agent:
         self.knowledge_retriever = knowledge_retriever
         self.references_format = references_format
 
+        self.skills = skills
+
         self.metadata = metadata
 
         self.tools = list(tools) if tools else []
@@ -2022,6 +2033,7 @@ class Agent:
                         run_response=run_response,
                         run_context=run_context,
                         session=agent_session,
+                        async_mode=True,
                     )
 
                     # 6. Prepare run messages
@@ -2369,6 +2381,7 @@ class Agent:
                         run_response=run_response,
                         run_context=run_context,
                         session=agent_session,
+                        async_mode=True,
                     )
 
                     # 6. Prepare run messages
@@ -3979,6 +3992,7 @@ class Agent:
                         run_response=run_response,
                         run_context=run_context,
                         session=agent_session,
+                        async_mode=True,
                     )
 
                     # 6. Prepare run messages
@@ -4285,6 +4299,7 @@ class Agent:
                         run_response=run_response,
                         run_context=run_context,
                         session=agent_session,
+                        async_mode=True,
                     )
 
                     # 6. Prepare run messages
@@ -6396,6 +6411,10 @@ class Agent:
             if self.update_knowledge:
                 agent_tools.append(self.add_to_knowledge)
 
+        # Add tools for accessing skills
+        if self.skills is not None:
+            agent_tools.extend(self.skills.get_tools())
+
         return agent_tools
 
     async def aget_tools(
@@ -6489,6 +6508,10 @@ class Agent:
             if self.update_knowledge:
                 agent_tools.append(self.add_to_knowledge)
 
+        # Add tools for accessing skills
+        if self.skills is not None:
+            agent_tools.extend(self.skills.get_tools())
+
         return agent_tools
 
     def _determine_tools_for_model(
@@ -6498,6 +6521,7 @@ class Agent:
         run_response: RunOutput,
         run_context: RunContext,
         session: AgentSession,
+        async_mode: bool = False,
     ) -> List[Union[Function, dict]]:
         _function_names = []
         _functions: List[Union[Function, dict]] = []
@@ -6528,7 +6552,8 @@ class Agent:
 
                 elif isinstance(tool, Toolkit):
                     # For each function in the toolkit and process entrypoint
-                    for name, _func in tool.functions.items():
+                    toolkit_functions = tool.get_async_functions() if async_mode else tool.get_functions()
+                    for name, _func in toolkit_functions.items():
                         if name in _function_names:
                             continue
                         _function_names.append(name)
@@ -6758,6 +6783,18 @@ class Agent:
         """
         return cancel_run_global(run_id)
 
+    @staticmethod
+    async def acancel_run(run_id: str) -> bool:
+        """Cancel a running agent execution (async version).
+
+        Args:
+            run_id (str): The run_id to cancel.
+
+        Returns:
+            bool: True if the run was found and marked for cancellation, False otherwise.
+        """
+        return await acancel_run_global(run_id)
+
     # -*- Session Database Functions
     def _read_session(
         self, session_id: str, session_type: SessionType = SessionType.AGENT
@@ -7171,7 +7208,6 @@ class Agent:
         """
         if self._has_async_db():
             raise ValueError("Async database not supported for save_session")
-
         # If the agent is a member of a team, do not save the session to the database
         if (
             self.db is not None
@@ -7937,6 +7973,11 @@ class Agent:
         # 3.3.8 Then add additional context
         if self.additional_context is not None:
             system_message_content += f"{self.additional_context}\n"
+        # 3.3.8.1 Then add skills to the system prompt
+        if self.skills is not None:
+            skills_snippet = self.skills.get_system_prompt_snippet()
+            if skills_snippet:
+                system_message_content += f"\n{skills_snippet}\n"
         # 3.3.9 Then add memories to the system prompt
         if self.add_memories_to_context:
             _memory_manager_not_set = False
@@ -8281,6 +8322,11 @@ class Agent:
         # 3.3.8 Then add additional context
         if self.additional_context is not None:
             system_message_content += f"{self.additional_context}\n"
+        # 3.3.8.1 Then add skills to the system prompt
+        if self.skills is not None:
+            skills_snippet = self.skills.get_system_prompt_snippet()
+            if skills_snippet:
+                system_message_content += f"\n{skills_snippet}\n"
         # 3.3.9 Then add memories to the system prompt
         if self.add_memories_to_context:
             _memory_manager_not_set = False