strands-agents/sdk-python

Issues

• [BUG] LiteLLM Moonshot AI Kimi K2 thinking fails on multi-turn conversations with 'Invalid part type: thinking'

  #1150 opened 2 days ago by westonbrown
  0

• LiteLLM: Moonshot AI rejects 'thinking' content type in multi-turn conversations

  #1149 opened 2 days ago by westonbrown
  1

• [FEATURE] Add all tool details under gen_ai.agent.tools attribute

  #1083 opened 2 days ago by kiranramnath007
  4

• [FEATURE] Live Prompt Adjustment

  #1130 opened 5 days ago by enitinb
  1

• [FEATURE] Tool Replacement

  #1131 opened 5 days ago by enitinb
  1

• [FEATURE] Add Agent latest_usage and latest_metrics besides `accumulated_usage` and `accumulated_metrics`

  #1144 opened 2 days ago by poshinchen
  0

• [FEATURE] Prompt caching support for all models

  #1140 opened 3 days ago by dbschmigelski
  0

• [FEATURE] Agent State Management - Snapshot, Pause, and Resume

  #1138 opened 4 days ago by nagabharann
  0

• [FEATURE] Visual Builder

  #1134 opened 5 days ago by enitinb
  1

• [BUG] With bedrock guardrails, tool output is redacted breaking the conversation

  #1077 opened 4 days ago by leotac
  1

• [BUG] Input/output message not redacted when guardrails_trace != "enabled"

  #1075 opened 4 days ago by leotac
  0

• [BUG] Decorator functions send the entire doc string as the description of the tool

  #1067 opened 18 days ago by zastrowm
  1

• [BUG] GeminiModel doesn't support the structured_output implementation

  #1121 opened 8 days ago by dpolistwm
  7

• [BUG] GeminiModel failure when using Vertex AI mode without tools

  #1039 opened 23 days ago by pshiko
  1

• [BUG][Graph] `structured_output_model` prevents correct output propagation from dependencies

  #1118 opened 8 days ago by courtem
  3

• [FEATURE] Tools and Agents event emit for websocket

  #1133 opened 5 days ago by enitinb
  0

• [FEATURE] Dynamic Role-Based Agent Reconfiguration

  #1132 opened 5 days ago by enitinb
  0

• [FEATURE] Re-try limit with structured_output

  #1108 opened 10 days ago by FlanaganSe
  6

• [FEATURE] Hook event for delta

  #1127 opened 7 days ago by tmokmss
  1

• [FEATURE] Implement Gemini Tool Choice

  #1129 opened 5 days ago by afarntrog
  0

• [FEATURE] Make invocation_state available as a parameter in the condition function of Graph Edge

  #1124 opened 8 days ago by schatterjee90
  0

• [BUG] Unssuported content type for "reasoning_content"

  #1056 opened 19 days ago by eliasto
  2

• [BUG] AgentResult.structured_output returns BaseModel | None

  #1114 opened 9 days ago by LeviLong01
  0

• [FEATURE] add MCP resource operations

  #1116 opened 9 days ago by xiehust
  0

• [FEATURE] Resume Session from ToolUse

  #1111 opened 10 days ago by Unshure
  0

• [FEATURE] Support a fixed, injected `AsyncOpenAI` client to enable alternative interface-compatible clients

  #1103 opened 11 days ago by dg-bain
  3

• [BUG] Structured output failing for LiteLLM model run with azure provider

  #1076 opened 10 days ago by piotrkubicki
  2

• [BUG] Invalid tool names result in dangling toolUse blocks

  #1069 opened 11 days ago by zastrowm
  3

• [BUG] `gen_ai.agent.name` OTel attribute used for MultiAgent spans even when an "agent" may not be used

  #1105 opened 11 days ago by dbschmigelski
  0

• [BUG] unable to install strands dependency

  #1102 opened 11 days ago by bsmahi
  0

• [FEATURE] The Bedrock documentation source can only accept bytes.

  #1066 opened 18 days ago by HawkClaws
  3

• [FEATURE] Bedrock Prompt Integration

  #1043 opened 23 days ago by linvald
  1

• [BUG] Lambda runtime error "Failed to load context: contextvars_context"

  #1086 opened 11 days ago by ari-luokkala
  3

• [BUG] property defaults to type string even if type is defined via anyOf

  #1101 opened 12 days ago by wistuba
  0

• [BUG] property defaults to type string even if type is defined via anyOf

  #1100 opened 12 days ago by wistuba
  0

• [BUG] Amazon Nova Models Hallucinate with JSON Content Blocks in Tool Results

  #1095 opened 13 days ago by maaajo
  0

• [BUG] Strands agent calls tool with parameter as dictionary instead of pydantic object

  #1090 opened 15 days ago by ruchikaandroid
  1

• [FEATURE]: Add Model-Agnostic Guardrails Documentation Using LiteLLM and Hooks

  #1088 opened 15 days ago by Arindam200
  0

• [FEATURE] Add Graph constructor option to toggle edge execution logic from OR to AND

  #1081 opened 17 days ago by dbschmigelski
  1

• [FEATURE] Improve handling for models not supporting reasoning content

  #1053 opened 20 days ago by Luke-Shepp
  1

• [FEATURE] Add Dapr statestore as session manager

  #1073 opened 17 days ago by filintod
  0

• [BUG] Strands Agent incorrectly sets SpanKind for Agent Spans

  #1044 opened 18 days ago by liustve
  2

• [FEATURE] Move Agent Json loading out of Experimental

  #1065 opened 18 days ago by Unshure
  0

• [FEATURE] Support Every model provider in Agent Json feature

  #1064 opened 18 days ago by Unshure
  0

• [BUG] Tool metrics (tool_trace) not recorded for interrupted tool calls

  #1063 opened 18 days ago by evehsu
  1

• [FEATURE] Add explicit context caching support for Gemini models

  #1060 opened 19 days ago by JosephGabito
  0

• [FEATURE] Add a configuration to disable OpenTelemetry instrumentation

  #1059 opened 19 days ago by B-Step62
  0

• [FEATURE] Support for Gemini Built-in Tools (GoogleSearch, CodeExecution, etc.)

  #1049 opened 21 days ago by pshiko
  0

• [BUG] OpenAI Implementation does not support passing custom HTTP client

  #1036 opened 25 days ago by karllessard
  7

• [FEATURE] Add Masking Function for Sensitive Data Before set_attribute

  #1033 opened 25 days ago by jjayp4rk
  0