Changelog

New Features:

• PPTX Reader: Added support for reading Microsoft PowerPoint (.pptx) files by
  implementing a dedicated PPTXReader class.
• Filter tool calls from History: Add ability to load fixed number of tool calls from history using max_tool_calls_from_history. Helps manage context size and reduce tokens. See the Docs for more information.

Improvements

• Workflow Cancellation: Improved the persistence of content when a cancellation happens.

Bug Fixes:

• Custom Events: Fixed edge cases related to Custom Events not being persisted to storage correctly.

What's Changed

• feat: add pptx reader support by @10Ala10 in #5022
• feat: Filter tool calls from history by @Mustafa-Esoofally in #5073
• fix: Metadata on teams for posthooks by @dirkbrnd in #5153
• fix: Custom Events by @dirkbrnd in #5161
• chore: persist workflow mid-run cancel data by @kausmeows in #5150
• chore: Release 2.2.1 by @dirkbrnd in #5163

Full Changelog: v2.2.0...v2.2.1

Changelog

Improvements:

• Update workflows arun to return AsyncIterator: The ****.arun method of Workflow now returns an async iterator, similar to Agent and Team. Examplein our docs.
• New Events on Agent/Team:
  • PostHookStarted, PostHookCompleted for both Agent and Team events
  • SessionSummaryCreationStarted, SessionSummaryCreationCompleted for both Agent and Team events
  • RunContentCompleted for both Agent and Team → This signals that the content stream has ended, which is useful for building UIs using the events
• Concurrent Memory Creation: Automatic memory creation is now started on a thread/task at the start of an Agent/Team run, which significantly improves runtime performance where memory generation is enabled.
• Get Run Output on Members: Improved get_run_output and get_last_run_output on Agent/Team to allow retrieval of runs from members. Therefore after a team has executed you can get the last run straight from the member agent.
• Team Delegation Updates:
  • Made various updates to the prompts and tool instructions for team delegation, to improve accuracy of member delegation.
  • Removed expected_output as a default parameter on the delegate_task_to_member tool call for Team. This is done to increase flexibility with teams and to avoid incorrect behaviour on the team leader.
• Team History Updates:
  • Adds add_team_history_to_members to share team-level request/response history with team members.
  • Improved how share_member_interactions work to make sharing interactions during a run more accurate.
  • Here are the related docs
• stream_events flag: Add a stream_events flag to Agent, Team, Workflow and all run methods to emit all events whens streaming a response.
• Session State Updating: Added update_session_state and aupdate_session_state functions on Agent and Team to allow direct updating of the session state that is persisted in the DB.
• AgentOS Session Endpoints: Added new endpoints:
  • POST /sessions → Create a new empty session (reference)
  • GET /sessions/{id}/runs/{id} → Get a run by its ID (reference)
  • PATCH /sessions/{id} → Update existing session (reference)

Bug Fixes:

• None values in Ollama metrics: Fix a problem calculating metrics with Ollama when certain values were None instead of 0.
• Only-media inputs: The Memory and Summary features now work with inputs containing only media
• Input media with Teams: Fix a problem with input media being ignored when streaming a Team’s response.

Breaking Changes and Deprecations

• Workflow.arun(..., stream=True) now return an AsyncIterator, i.e. which means await is not required, but instead async for event in workflow.arun(...) is the correct syntax.
• All events besides RunContent events (the actual model content chunks) are now behind the stream_events flag and will only be yielded where stream_events=True . Previously tool-call events were yielded regardless.
• stream_intermediate_steps has been deprecated in favour of stream_events

What's Changed

• fix: Use run_input media in Team._arun_stream by @praveen-livspace in #5127
• fix: Handle None values in Ollama metrics calculation by @ron-42 in #5085
• fix: accept inputs containing only media by @kausmeows in #5129
• feat: Add instantiation time to /health by @dirkbrnd in #4992
• fix: Concurrent memories, session summary events, run content completion events by @dirkbrnd in #4998
• chore: workflows arun returns asynciterator by @kausmeows in #5100
• feat: add stream_events param everywhere by @manuhortet in #5092
• fix: Various issues with team history by @dirkbrnd in #5104
• feat: New session endpoints by @dirkbrnd in #5133
• test: add unit test for pandas tool by @nhankiet in #5140
• chore: Release 2.2.0 by @dirkbrnd in #5146

New Contributors

• @praveen-livspace made their first contribution in #5127
• @ron-42 made their first contribution in #5085
• @nhankiet made their first contribution in #5140

Full Changelog: v2.1.10...v2.2.0

Changelog

New Features:

• [Experimental feature] Culture: You can now provide your agents with a shared space to think, write, and build on each other's ideas. This will give your agents the ability to learn collectively.
• Gmail tools: ****Mark as read, mark as unread now supported in gmail toolkit.
• Knowledge on AgentOS: Knowledge can now be added and managed via any AgentOS interface without needing to attach it to an Agent or Team.

What's Changed

• feat: Add gmail tools mark-as-read function by @willemcdejongh in #4390
• feat: Support knowledge on agentOS without an agent or team by @willemcdejongh in #5115
• Fix: formatting by @ashpreetbedi in #5130
• feat: introduce culture by @manuhortet in #5124
• chore: v2.1.10 by @manuhortet in #5132

Full Changelog: v2.1.9...v2.1.10

Improvements:

• id on Message: Add id field on the Message class that is used internally for messages sent to LLMs, but also for the list of messages on the RunOutput objects. This can be used for tracking messages in storage.
• Extend support for session_state in Workflows: Provide access to the contextual session_state to the evaluator and selector functions of the Condition and Router Workflow Step classes.

Bug Fixes:

• Metadata On Teams: Fixed issues with passing metadata on run and arun for teams.
• Reader on async_add_contents: Fixed missing reader attribute not being passed to forknowledge loader functions

What's Changed

• Fix: formatting by @ashpreetbedi in #5109
• fix: Metadata on teams by @dirkbrnd in #5108
• fix: Async_add_contents reader by @willemcdejongh in #5043
• feat: Add id on Message by @dirkbrnd in #5110
• feat: propogate session_state to condition/router evaluator function by @kausmeows in #5107
• chore: v2.1.9 by @manuhortet in #5122

Full Changelog: v2.1.8...v2.1.9

Changelog

New Features:

• Class-based executors in Workflows: Enables to use a class-based executor by defining a class that implements the __call__ method. See docs
• Post-hook support on streaming: adds support for post-hooks on streaming flows.

Improvements:

• Update GoogleCalenderTools to Notify attendees when creating,updating or deleting event in calendar
• Update Knowledge search endpoint naming: update endpoint name from search_vectors to search_knowledge
• Jira Toolkit: Add logic to the JiraTools toolkit to enable creating worklogs.

Bug Fixes:

• Fix AttributeError when continuing agent run with run_id only: Fix AttributeError that occurs when calling acontinue_run() or acontinue_run_stream() methods with only run_id parameter and run_response=None. The logging statements were trying to access run_response.run_id when run_response could be None.
• Fix Dependencies not being resolved in Agent Async run: update to pass the correct dependencies variable name in agent async run

What's Changed

• Fix: fix AttributeError when continuing agent run with run_id only by @10Ala10 in #5070
• chore: enhance cookbook demo with more examples by @uzaxirr in #5012
• feat: Notify attendees when creating,updating or deleting event in calendar by @srexrg in #4910
• fix: pass the correct dependencies variable name in agent async run by @harrytran001 in #5080
• chore: Update Knowledge search endpoint naming by @willemcdejongh in #5088
• chore: Handle class-based asynchronous executors by @benmezger in #4904
• fix: update response_audio type in schemas to a single dict by @anuragts in #5087
• feat: Add post-hook support on streaming by @dirkbrnd in #5068
• feat: extend Jira toolkit to allow adding worklogs by @rohitnishant in #4991
• chore: Release 2.1.8 by @kausmeows in #5090

New Contributors

• @benmezger made their first contribution in #4904
• @rohitnishant made their first contribution in #4991

Full Changelog: v2.1.7...v2.1.8

Changelog

Bug Fixes:

• Memory with async databases: Fixed an error updating user memories when using an async database implementation, like PostgresDb.

What's Changed

• fix: enhance memory retrieval for asynchronous database support by @JasonLovesDoggo in #5067
• chore: async postgres imports by @manuhortet in #5069
• chore: Release 2.1.7 by @ysolanky in #5074

New Contributors

• @JasonLovesDoggo made their first contribution in #5067

Full Changelog: v2.1.6...v2.1.7

Changelog

New features

• Support for SurrealDB: Add complete support to use a SurrealDB database with your Agents, Teams and Workflows, via the new SurrealDb class.

Improvements:

• User Memory: Updated some prompts to increase general accuracy at finding relevance user preferences to keep as user memories.
• V2 migration: Empty updated_at values can now be preserved if requested when using our v2 migration script.
• Updates to store_tool_results flag:
  • Renamed it to store_tool_messages for better clarity
  • Tool message pairs are removed together: To maintain valid message sequences, when tool messages are removed from storage, the corresponding assistant messages that contains tool calls for that tool message are also removed. Most model providers require a strict "tool call pair"
  • Metrics still include token counts: Your run metrics will still reflect the actual tokens used during execution, including the tool calls and results. Only the message storage is affected.

What's Changed

• fix: Cookbook file url by @Mustafa-Esoofally in #5057
• [feat] Add SurrealDb by @martinschaer in #4575
• feat: memory management for large v2 migrations by @manuhortet in #5021
• fix: Clean up unnecessary attributes by @dirkbrnd in #5061
• feat: adjust main memory prompts by @manuhortet in #5064
• chore: implementation update to store_history_messages and store_tool_messages by @kausmeows in #5053
• chore: release v2.1.6 by @manuhortet in #5063

New Contributors

• @martinschaer made their first contribution in #4575

Full Changelog: v2.1.5...v2.1.6

Changelog

New Features:

• Async Postgres (docs): Added support for async Postgres all across the library.
• Add flags to disable storing tool and history messages on run output: Agent and Team now have store_history_messages and store_tool_messages to control whether these extra messages should be stored or not.
• Add stream_executor_events to filter workflow executor events (docs):
  • using stream_intermediate_events we control to switch on-off the workflow events like WorkflowStarted, StepStarted etc.
  • to filter out events coming from agent/team/custom function we have this stream_executor_events now similar to stream_member_events in Team.
• Knowledge search endpoints (docs): Allow searching vectorDB via AgentOS API.
• Reasoning models: added support for Gemini 2.5+, Anthropic Claude, and VertexAI Claude native thinking capabilities when used as reasoning models with your Agents.

Improvements:

• AgentOS Access Logs: Disabled access logs for AgentOS by default. You can enable by setting serve(access_log=True).
• GoogleSheets Service Accounts: Add support for service account auth on GoogleSheetsTools.
• PineconeDB async upsert logic: Add support for the content hash and metadata fields when using the async upsert method of the PineconeDb implementation.
• MCP toolbox for databases: Allow MCP server urls not containing the conventional /mcp segment.
• User memory creation in async scenarios: Track agent_id or team_id when creating memories in all async scenarios.

Bug Fixes:

• AG-UI text message after Tool calls: Fixed a problem handling tool calls and message events sequentially in the AG-UI interface.
• Metrics run counts with SQLiteDb and Firestore: Fixed a bug where wrong metrics would be returned for the Run count field when using the SQLiteDb or FirestoreDb implementations.
• V2 database migration script:
  • The updated_at field is now preserved when migrating sessions and memories.
  • Removed some unnecessary dependencies, like pymongo.

What's Changed

• feat: Disable access log for AgentOS by @dirkbrnd in #4971
• fix: sqlite db correct runs count in metrics calculation by @Ayush0054 in #4990
• [fix] AGUI - message after tool call cause errors by @Mustafa-Esoofally in #4972
• chore: SQLiteDB initialization DX improvement by @uzaxirr in #5007
• fix: Fix missing content hash deletion and metadata in PineconeDB async_upsert by @10Ala10 in #5006
• fix: Knowledge docx handling and error status loading by @willemcdejongh in #5025
• [feat] Add support for service accounts in the Google Sheets tool. by @simonkagwi in #4997
• fix: fix PDF reader chunking strategy initialization bug by @10Ala10 in #5023
• Fix: Variable 'result' is referenced before assignment in Get Knowledge Contents MySQLDb by @harrytran001 in #5019
• feat: stop enforcing /mcp in urls used with MCPToolbox by @manuhortet in #5028
• feat: add flags to disable storing tool and history messages on run output by @kausmeows in #5029
• feat: Search knowledge API by @willemcdejongh in #4989
• fix: Remove render_heavy_js argument from ScrapeGraphTools.searchscraper by @k3vob in #5031
• chore: Update KB vector search endpoint name by @willemcdejongh in #5041
• feat: Extract content middleware by @dirkbrnd in #5042
• chore: stream_executor_events to filter workflow executor events by @kausmeows in #5038
• chore: AgentOS interfaces cookbook by @willemcdejongh in #5044
• chore: update get_run_input logic by @kausmeows in #5035
• fix: preserve updated_at fields when migrating to v2 by @manuhortet in #5046
• fix: get media on top level on GET runs api by @anuragts in #4936
• feat: Add reasoning support for Gemini, Anthropic, and VertexAI Claude models by @uzaxirr in #5009
• fix: remove unnecessary dependencies from the v2 migration logic by @manuhortet in #5051
• fix: add missing team_id and agent_id when creating memory with the async methods by @kepler in #4895
• fix: input typo by @anuragts in #5049
• chore: Update style checks in CI by @willemcdejongh in #5027
• feat: support for async databases + async postgres implementation by @manuhortet in #4338
• chore: release v2.1.5 by @manuhortet in #5052

New Contributors

• @simonkagwi made their first contribution in #4997
• @k3vob made their first contribution in #5031

Full Changelog: v2.1.4...v2.1.5

Changelog

New Features:

• Workflow History: Support to add workflow history to all or individual steps for a better continuous conversational experience
• Google Drive Toolkit: Added the GoogleDriveTools class, to provide your Agents with read and write access to Google Drive.

Improvements:

• Update parallel and custom function step events streaming logic in workflows:
  • Inspect all events that are yielded during the workflow run, and set the workflow context variables on them before yielding upwards.
  • Update parallel step event streaming logic to immediately yield events as they come so that FE can catch and display in real time.
    • Old approach - collect all, yield at end
    • New approach - stream immediately (used queues as we cant yield directly from async running tasks)
• Custom events with AG-UI: The AG-UI integration now support custom events. Your Agno custom events will now be delivered in the AG-UI interface in the conventional AG-UI custom event format.
• Tool post-hooks in case of tool failure: Our tool post-hooks now also run when the tool run fails. This enables you to easily add logic to run in case of tool failures!

Bug Fixes:

• Order of persisted streaming events: Fix a bug where the created_at of some Memory events would be larger than the created_at of the final RunCompleted event.
• Handle Team sessions with no runs: When cancelling a Team run early enough, you can end with a session that contains no run objects. Some functionality was failing in this edge case.
• Dynamo Table definitions: Fixed a bug where some DynamoDB schemas incorrectly defined non key attributes
• Dictionary Key Order Causing Redundant Cache Files: The current caching in agno/tools/function.py used str() to serialize dictionary arguments for cache key generation. Since str() preserves dictionary key order, identical data with different key orders produced different cache keys. Replaced str() with json.dumps(sort_keys=True) to ensure serialization regardless of dictionary key order
• Change dimensions type to Optional[int] in FastEmbedEmbedder : Fixed an issue where FastEmbedEmbedder didn’t allow optional dimensions.

What's Changed

• [feat] Add support for custom events in AGUI interface by @kredde in #4960
• [Fix]: Remove unused from and add missing fields into schemas in DynamoDB + fix Dynamo Serialization and Query Syntax by @harrytran001 in #4771
• Fix: Dictionary Key Order Causing Redundant Cache Files by @uzaxirr in #4907
• fix: handle team sessions with no runs by @manuhortet in #4951
• fix: ensure the RunCompletedEvent is the last created event by @manuhortet in #4959
• chore: customer research workflow agent os by @kausmeows in #4950
• chore: deactivate CI AWS tests by @manuhortet in #4969
• chore: update parallel and custom function step events streaming logic by @kausmeows in #4954
• fix: change dimensions type to Optional[int] in FastEmbedEmbedder by @Ayush0054 in #4981
• chore: cancel previous workflow runs on new commits by @manuhortet in #4980
• feat: add Google Drive toolkit by @SaiPoornaChandraPrakash in #4658
• fix: Run tool post hooks when tools fail with exceptions by @jtalmi in #4926
• chore: store events for agent os in workflows by @kausmeows in #4982
• fix: add events to the OS workflow run schema by @manuhortet in #4986
• feat: workflow history by @kausmeows in #4785
• chore: Release 2.1.4 by @kausmeows in #4966

New Contributors

• @kredde made their first contribution in #4960
• @harrytran001 made their first contribution in #4771
• @SaiPoornaChandraPrakash made their first contribution in #4658

Full Changelog: v2.1.3...v2.1.4

Changelog

New Features:

• Claude Vertex AI: Implemented a new Claude model class to be served through Vertex AI.
• Dynamic Model Router in OpenRouter: Added the ability to define multiple models in the OpenRouter Class in case the primary model fails

Improvements:

• AgentOS MCP server with custom FastAPI apps: AgentOS instances running with a custom FastAPI app as base app can now be exposed as MCP servers, using the enable_mcp_server parameter.
• Support MCP tools inside the AgentOS MCP server: The Agents, Teams and Workflows exposed via MCP in your AgentOS MCP server can now use MCP tools.
• FileTools with relative paths: Update the FileTools Toolkit to better handle using relative paths in all its methods.

Bug Fixes:

• Support MCP tools with custom FastAPI apps: Fixed a bug provoking errors when connecting to MCP servers inside an AgentOS setup with a custom FastAPI app as base app.
• Reading sessions with no runs: Fixed a bug reading Team sessions with no runs, which happens when cancelling a Team run early.
• Cancelling Workflow runs started via websocket: Fixed a bug that made run cancellation fail for workflow runs started via websocket.

What's Changed

• fix: workflow cancel flow by @kausmeows in #4947
• feat: better handle of MCP lifecycle in the AgentOS by @manuhortet in #4949
• feat: claude vertexai by @ysolanky in #4943
• fix: make file names relative to base_dir by @mickvav in #4836
• fixes #4611 - Added Fallback models in OpenRouter by @uzaxirr in #4906
• docs: add Opik telemetry docs by @vincentkoc in #4881
• chore: Release 2.1.3 by @ysolanky in #4953

New Contributors

• @mickvav made their first contribution in #4836
• @uzaxirr made their first contribution in #4906
• @vincentkoc made their first contribution in #4881

Full Changelog: v2.1.2...v2.1.3