agno 2.4.6__tar.gz → 2.4.8__tar.gz

{agno-2.4.6 → agno-2.4.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agno
-Version: 2.4.6
+Version: 2.4.8
 Summary: Agno: a lightweight library for building Multi-Agent Systems
 Author-email: Ashpreet Bedi <ashpreet@agno.com>
 Project-URL: homepage, https://agno.com
@@ -107,6 +107,8 @@ Requires-Dist: portkey-ai; extra == "portkey"
 Provides-Extra: tokenizers
 Requires-Dist: tiktoken; extra == "tokenizers"
 Requires-Dist: tokenizers; extra == "tokenizers"
+Provides-Extra: cel
+Requires-Dist: cel-python; extra == "cel"
 Provides-Extra: agentql
 Requires-Dist: agentql; extra == "agentql"
 Provides-Extra: apify
@@ -225,7 +227,7 @@ Requires-Dist: pgvector; extra == "pgvector"
 Provides-Extra: chromadb
 Requires-Dist: chromadb; extra == "chromadb"
 Provides-Extra: lancedb
-Requires-Dist: lancedb==0.24.0; extra == "lancedb"
+Requires-Dist: lancedb>=0.26.0; extra == "lancedb"
 Requires-Dist: tantivy; extra == "lancedb"
 Provides-Extra: pylance
 Requires-Dist: pylance; extra == "pylance"
@@ -445,7 +447,7 @@ Dynamic: license-file
 </div>
 
 <p align="center">
-  Build agents that learn.
+  Build multi-agent systems that learn.
 </p>
 
 <div align="center">
@@ -460,15 +462,16 @@ Dynamic: license-file
 
 ## What is Agno?
 
-**A Python SDK for building agents that learn and improve with every interaction.**
+**A framework for building multi-agent systems that learn and improve with every interaction.**
 
 Most agents are stateless. They reason, respond, forget. Session history helps, but they're exactly as capable on day 1000 as they were on day 1.
 
-Agno agents are different. They remember users across sessions, accumulate knowledge across conversations, and learn from decisions. Insights from one user benefit everyone.
+Agno agents are different. They remember users across sessions, accumulate knowledge across conversations, and learn from decisions. Insights from one user benefit everyone. The system gets smarter over time.
 
 Everything runs in your cloud. Your data never leaves your environment.
 
 ## Quick Example
+
 ```python
 from agno.agent import Agent
 from agno.db.sqlite import SqliteDb
@@ -485,46 +488,31 @@ One line. Your agent now remembers users, accumulates knowledge, and improves ov
 
 ## Production Stack
 
+Agno provides the complete infrastructure for building multi-agent systems that learn:
+
 | Layer | What it does |
 |-------|--------------|
-| **SDK** | Build agents with learning, tools, knowledge, and guardrails |
+| **Framework** | Build agents with learning, tools, knowledge, and guardrails |
 | **Runtime** | Run in production using [AgentOS](https://docs.agno.com/agent-os/introduction) |
 | **Control Plane** | Monitor and manage via the [AgentOS UI](https://os.agno.com) |
 
-## Features
-
-**Learning**
-- User profiles that persist across sessions
-- User memories that accumulate over time
-- Learned knowledge that transfers across users
-- Always or agentic learning modes
+## Get Started
 
-**Core**
-- Model-agnostic: OpenAI, Anthropic, Google, local models
-- Type-safe I/O with `input_schema` and `output_schema`
-- Async-first, built for long-running tasks
-- Natively multimodal (text, images, audio, video, files)
+1. [Build your first agent](https://docs.agno.com/first-agent)
+2. [Build your first multi-agent system](https://docs.agno.com/first-multi-agent-system)
+3. [Deploy to production](https://docs.agno.com/production/overview)
 
-**Knowledge**
-- Agentic RAG with 20+ vector stores, hybrid search, reranking
-- Persistent storage for session history and state
+More: [Docs](https://docs.agno.com) · [Cookbook](https://github.com/agno-agi/agno/tree/main/cookbook)
 
-**Orchestration**
-- Human-in-the-loop (confirmations, approvals, overrides)
-- Guardrails for validation and security
-- First-class MCP and A2A support
-- 100+ built-in toolkits
-
-**Production**
-- Ready-to-use FastAPI runtime
-- Integrated control plane UI
-- Evals for accuracy, performance, latency
-
-## Getting Started
+## Features
 
-1. Follow the [quickstart](https://docs.agno.com/get-started/quickstart)
-2. Browse the [cookbook](https://github.com/agno-agi/agno/tree/main/cookbook)
-3. Read the [docs](https://docs.agno.com)
+| Category | What you get |
+|----------|--------------|
+| **Learning** | User profiles that persist across sessions. User memories that accumulate over time. Learned knowledge that transfers across users. Always or agentic learning modes. |
+| **Core** | Model-agnostic: OpenAI, Anthropic, Google, local models. Type-safe I/O with `input_schema` and `output_schema`. Async-first, built for long-running tasks. Natively multimodal (text, images, audio, video, files). |
+| **Knowledge** | Agentic RAG with 20+ vector stores, hybrid search, reranking. Persistent storage for session history and state. |
+| **Orchestration** | Human-in-the-loop (confirmations, approvals, overrides). Guardrails for validation and security. First-class MCP and A2A support. 100+ built-in toolkits. |
+| **Production** | Ready-to-use FastAPI runtime. Integrated control plane UI. Evals for accuracy, performance, latency. |
 
 ## IDE Integration
 
@@ -536,7 +524,7 @@ Also works with VSCode, Windsurf, and similar tools.
 
 ## Contributing
 
-See the [contributing guide](https://github.com/agno-agi/agno/blob/v2.0/CONTRIBUTING.md).
+See the [contributing guide](https://github.com/agno-agi/agno/blob/main/CONTRIBUTING.md).
 
 ## Telemetry
 

{agno-2.4.6 → agno-2.4.8}/README.md

@@ -9,7 +9,7 @@
 </div>
 
 <p align="center">
-  Build agents that learn.
+  Build multi-agent systems that learn.
 </p>
 
 <div align="center">
@@ -24,15 +24,16 @@
 
 ## What is Agno?
 
-**A Python SDK for building agents that learn and improve with every interaction.**
+**A framework for building multi-agent systems that learn and improve with every interaction.**
 
 Most agents are stateless. They reason, respond, forget. Session history helps, but they're exactly as capable on day 1000 as they were on day 1.
 
-Agno agents are different. They remember users across sessions, accumulate knowledge across conversations, and learn from decisions. Insights from one user benefit everyone.
+Agno agents are different. They remember users across sessions, accumulate knowledge across conversations, and learn from decisions. Insights from one user benefit everyone. The system gets smarter over time.
 
 Everything runs in your cloud. Your data never leaves your environment.
 
 ## Quick Example
+
 ```python
 from agno.agent import Agent
 from agno.db.sqlite import SqliteDb
@@ -49,46 +50,31 @@ One line. Your agent now remembers users, accumulates knowledge, and improves ov
 
 ## Production Stack
 
+Agno provides the complete infrastructure for building multi-agent systems that learn:
+
 | Layer | What it does |
 |-------|--------------|
-| **SDK** | Build agents with learning, tools, knowledge, and guardrails |
+| **Framework** | Build agents with learning, tools, knowledge, and guardrails |
 | **Runtime** | Run in production using [AgentOS](https://docs.agno.com/agent-os/introduction) |
 | **Control Plane** | Monitor and manage via the [AgentOS UI](https://os.agno.com) |
 
-## Features
-
-**Learning**
-- User profiles that persist across sessions
-- User memories that accumulate over time
-- Learned knowledge that transfers across users
-- Always or agentic learning modes
+## Get Started
 
-**Core**
-- Model-agnostic: OpenAI, Anthropic, Google, local models
-- Type-safe I/O with `input_schema` and `output_schema`
-- Async-first, built for long-running tasks
-- Natively multimodal (text, images, audio, video, files)
+1. [Build your first agent](https://docs.agno.com/first-agent)
+2. [Build your first multi-agent system](https://docs.agno.com/first-multi-agent-system)
+3. [Deploy to production](https://docs.agno.com/production/overview)
 
-**Knowledge**
-- Agentic RAG with 20+ vector stores, hybrid search, reranking
-- Persistent storage for session history and state
+More: [Docs](https://docs.agno.com) · [Cookbook](https://github.com/agno-agi/agno/tree/main/cookbook)
 
-**Orchestration**
-- Human-in-the-loop (confirmations, approvals, overrides)
-- Guardrails for validation and security
-- First-class MCP and A2A support
-- 100+ built-in toolkits
-
-**Production**
-- Ready-to-use FastAPI runtime
-- Integrated control plane UI
-- Evals for accuracy, performance, latency
-
-## Getting Started
+## Features
 
-1. Follow the [quickstart](https://docs.agno.com/get-started/quickstart)
-2. Browse the [cookbook](https://github.com/agno-agi/agno/tree/main/cookbook)
-3. Read the [docs](https://docs.agno.com)
+| Category | What you get |
+|----------|--------------|
+| **Learning** | User profiles that persist across sessions. User memories that accumulate over time. Learned knowledge that transfers across users. Always or agentic learning modes. |
+| **Core** | Model-agnostic: OpenAI, Anthropic, Google, local models. Type-safe I/O with `input_schema` and `output_schema`. Async-first, built for long-running tasks. Natively multimodal (text, images, audio, video, files). |
+| **Knowledge** | Agentic RAG with 20+ vector stores, hybrid search, reranking. Persistent storage for session history and state. |
+| **Orchestration** | Human-in-the-loop (confirmations, approvals, overrides). Guardrails for validation and security. First-class MCP and A2A support. 100+ built-in toolkits. |
+| **Production** | Ready-to-use FastAPI runtime. Integrated control plane UI. Evals for accuracy, performance, latency. |
 
 ## IDE Integration
 
@@ -100,7 +86,7 @@ Also works with VSCode, Windsurf, and similar tools.
 
 ## Contributing
 
-See the [contributing guide](https://github.com/agno-agi/agno/blob/v2.0/CONTRIBUTING.md).
+See the [contributing guide](https://github.com/agno-agi/agno/blob/main/CONTRIBUTING.md).
 
 ## Telemetry
 

{agno-2.4.6 → agno-2.4.8}/agno/agent/agent.py

@@ -11493,7 +11493,7 @@ class Agent:
 
         return get_tool_call_history
 
-    def _update_session_state_tool(self, session_state, session_state_updates: dict) -> str:
+    def _update_session_state_tool(self, run_context: RunContext, session_state_updates: dict) -> str:
         """
         Update the shared session state.  Provide any updates as a dictionary of key-value pairs.
         Example:
@@ -11502,6 +11502,10 @@ class Agent:
         Args:
             session_state_updates (dict): The updates to apply to the shared session state. Should be a dictionary of key-value pairs.
         """
+
+        if run_context.session_state is None:
+            run_context.session_state = {}
+        session_state = run_context.session_state
         for key, value in session_state_updates.items():
             session_state[key] = value
 

{agno-2.4.6 → agno-2.4.8}/agno/db/base.py

@@ -576,6 +576,7 @@ class BaseDb(ABC):
         component_type: Optional[ComponentType] = None,
         name: Optional[str] = None,
         description: Optional[str] = None,
+        current_version: Optional[int] = None,
         metadata: Optional[Dict[str, Any]] = None,
     ) -> Dict[str, Any]:
         """Create or update a component.
@@ -585,6 +586,7 @@ class BaseDb(ABC):
             component_type: Type (agent|team|workflow). Required for create, optional for update.
             name: Display name.
             description: Optional description.
+            current_version: Optional current version.
             metadata: Optional metadata dict.
 
         Returns:

{agno-2.4.6 → agno-2.4.8}/agno/db/postgres/postgres.py

@@ -3241,6 +3241,7 @@ class PostgresDb(BaseDb):
         component_type: Optional[ComponentType] = None,
         name: Optional[str] = None,
         description: Optional[str] = None,
+        current_version: Optional[int] = None,
         metadata: Optional[Dict[str, Any]] = None,
     ) -> Dict[str, Any]:
         """Create or update a component.
@@ -3250,6 +3251,7 @@ class PostgresDb(BaseDb):
             component_type: Type (agent|team|workflow). Required for create, optional for update.
             name: Display name.
             description: Optional description.
+            current_version: Optional current version.
             metadata: Optional metadata dict.
 
         Returns:
@@ -3317,6 +3319,8 @@ class PostgresDb(BaseDb):
                         updates["name"] = name
                     if description is not None:
                         updates["description"] = description
+                    if current_version is not None:
+                        updates["current_version"] = current_version
                     if metadata is not None:
                         updates["metadata"] = metadata
 
@@ -3693,8 +3697,8 @@ class PostgresDb(BaseDb):
             raise ValueError(f"Invalid stage: {stage}")
 
         try:
-            configs_table = self._get_table(table_type="component_configs", create_table_if_not_found=True)
             components_table = self._get_table(table_type="components")
+            configs_table = self._get_table(table_type="component_configs", create_table_if_not_found=True)
             links_table = self._get_table(table_type="component_links", create_table_if_not_found=True)
 
             if components_table is None:
@@ -3881,10 +3885,6 @@ class PostgresDb(BaseDb):
                 if config_row is None:
                     return False
 
-                # Cannot delete published configs
-                if config_row == "published":
-                    raise ValueError(f"Cannot delete published config {component_id} v{version}")
-
                 # Check if it's current version
                 current = sess.execute(
                     select(components_table.c.current_version).where(components_table.c.component_id == component_id)

{agno-2.4.6 → agno-2.4.8}/agno/db/singlestore/singlestore.py

@@ -173,8 +173,7 @@ class SingleStoreDb(BaseDb):
                     column_kwargs["unique"] = True
                 columns.append(Column(*column_args, **column_kwargs))
 
-            # Create the table object without constraints to avoid autoload issues
-            table = Table(table_name, self.metadata, *columns, schema=self.db_schema)
+            table = Table(table_name, self.metadata, *columns, schema=self.db_schema, extend_existing=True)
 
             return table
 
@@ -240,8 +239,7 @@ class SingleStoreDb(BaseDb):
 
                 columns.append(Column(*column_args, **column_kwargs))
 
-            # Create the table object
-            table = Table(table_name, self.metadata, *columns, schema=self.db_schema)
+            table = Table(table_name, self.metadata, *columns, schema=self.db_schema, extend_existing=True)
 
             # Add multi-column unique constraints with table-specific names
             for constraint in schema_unique_constraints:
@@ -396,7 +394,8 @@ class SingleStoreDb(BaseDb):
 
         if table_type == "spans":
             # Ensure traces table exists first (for foreign key)
-            self._get_table(table_type="traces", create_table_if_not_found=create_table_if_not_found)
+            if create_table_if_not_found:
+                self._get_table(table_type="traces", create_table_if_not_found=True)
             self.spans_table = self._get_or_create_table(
                 table_name=self.span_table_name,
                 table_type="spans",

{agno-2.4.6 → agno-2.4.8}/agno/db/sqlite/sqlite.py

@@ -3139,6 +3139,7 @@ class SqliteDb(BaseDb):
         component_type: Optional[ComponentType] = None,
         name: Optional[str] = None,
         description: Optional[str] = None,
+        current_version: Optional[int] = None,
         metadata: Optional[Dict[str, Any]] = None,
     ) -> Dict[str, Any]:
         """Create or update a component.
@@ -3148,6 +3149,7 @@ class SqliteDb(BaseDb):
             component_type: Type (agent|team|workflow). Required for create, optional for update.
             name: Display name.
             description: Optional description.
+            current_version: Optional current version.
             metadata: Optional metadata dict.
 
         Returns:
@@ -3213,6 +3215,8 @@ class SqliteDb(BaseDb):
                         updates["name"] = name
                     if description is not None:
                         updates["description"] = description
+                    if current_version is not None:
+                        updates["current_version"] = current_version
                     if metadata is not None:
                         updates["metadata"] = metadata
 
@@ -3759,10 +3763,6 @@ class SqliteDb(BaseDb):
                 if config_row is None:
                     return False
 
-                # Cannot delete published configs
-                if config_row.stage == "published":
-                    raise ValueError(f"Cannot delete published config {component_id} v{version}")
-
                 # Check if it's current version
                 current = sess.execute(
                     select(components_table.c.current_version).where(components_table.c.component_id == component_id)