Adjust docs n tests for better model defaults

Signed-off-by: Christopher Petito <chrisjpetito@gmail.com>

Author: krissetto

cagent-schema.json

@@ -451,15 +451,15 @@
         },
         "provider_opts": {
           "type": "object",
-          "description": "Provider-specific options. dmr: runtime_flags. anthropic: interleaved_thinking (boolean, default false). openai/anthropic/google: rerank_prompt (string) to fully override the system prompt used for RAG reranking (advanced - prefer using results.reranking.criteria for domain-specific guidance).",
+          "description": "Provider-specific options. dmr: runtime_flags. anthropic/amazon-bedrock (Claude): interleaved_thinking (boolean, default true). openai/anthropic/google: rerank_prompt (string) to fully override the system prompt used for RAG reranking (advanced - prefer using results.reranking.criteria for domain-specific guidance).",
           "additionalProperties": true
         },
         "track_usage": {
           "type": "boolean",
           "description": "Whether to track usage"
         },
         "thinking_budget": {
-          "description": "Controls reasoning effort/budget. OpenAI: string levels ('minimal','low','medium','high'). Anthropic: integer token budget (1024-32768). Gemini: integer token budget (-1 for unlimited, 0 to disable, 24576 max).",
+          "description": "Controls reasoning effort/budget. OpenAI: string levels ('minimal','low','medium','high'), default 'medium'. Anthropic: integer token budget (1024-32768), default 8192. Amazon Bedrock (Claude): same as Anthropic. Google Gemini 2.5: integer token budget (-1 for dynamic, 0 to disable, 24576 max), default -1. Google Gemini 3: string levels ('minimal' Flash only,'low','medium','high'), default 'high' for Pro, 'medium' for Flash.",
           "oneOf": [
             {
               "type": "string",
@@ -469,21 +469,23 @@
                 "medium",
                 "high"
               ],
-              "description": "Reasoning effort level (OpenAI)"
+              "description": "Reasoning effort level (OpenAI, Gemini 3)"
             },
             {
               "type": "integer",
               "minimum": -1,
               "maximum": 32768,
-              "description": "Token budget for extended thinking (Anthropic, Google)"
+              "description": "Token budget for extended thinking (Anthropic, Bedrock Claude, Gemini 2.5)"
             }
           ],
           "examples": [
             "minimal",
             "low",
             "medium",
             "high",
+            -1,
             1024,
+            8192,
             32768
           ]
         },

e2e/cagent_exec_test.go

@@ -51,31 +51,48 @@ func TestExec_OpenAI_gpt5_codex(t *testing.T) {
 func TestExec_Anthropic(t *testing.T) {
 	out := cagentExec(t, "testdata/basic.yaml", "--model=anthropic/claude-sonnet-4-0", "What's 2+2?")
 
-	require.Equal(t, "\n--- Agent: root ---\n2 + 2 = 4", out)
+	// With interleaved thinking enabled by default, Anthropic responses include thinking content
+	require.Contains(t, out, "--- Agent: root ---")
+	require.Contains(t, out, "2 + 2 = 4")
 }
 
 func TestExec_Anthropic_ToolCall(t *testing.T) {
 	out := cagentExec(t, "testdata/fs_tools.yaml", "--model=anthropic/claude-sonnet-4-0", "How many files in testdata/working_dir? Only output the number.")
 
-	require.Equal(t, "\n--- Agent: root ---\n\nCalling list_directory(path: \"testdata/working_dir\")\n\nlist_directory response → \"FILE README.me\\n\"\n1", out)
+	// With interleaved thinking enabled by default, Anthropic responses include thinking content
+	require.Contains(t, out, "--- Agent: root ---")
+	require.Contains(t, out, `Calling list_directory(path: "testdata/working_dir")`)
+	require.Contains(t, out, `list_directory response → "FILE README.me\n"`)
+	// The response should end with "1" (the count)
+	require.True(t, out != "" && out[len(out)-1] == '1', "response should end with '1'")
 }
 
 func TestExec_Anthropic_AgentsMd(t *testing.T) {
 	out := cagentExec(t, "testdata/agents-md.yaml", "--model=anthropic/claude-sonnet-4-0", "What's 2+2?")
 
-	require.Equal(t, "\n--- Agent: root ---\n2 + 2 = 4", out)
+	// With interleaved thinking enabled by default, Anthropic responses include thinking content
+	require.Contains(t, out, "--- Agent: root ---")
+	require.Contains(t, out, "2 + 2 = 4")
 }
 
 func TestExec_Gemini(t *testing.T) {
 	out := cagentExec(t, "testdata/basic.yaml", "--model=google/gemini-2.5-flash", "What's 2+2?")
 
-	require.Equal(t, "\n--- Agent: root ---\n2 + 2 = 4", out)
+	// With thinking enabled by default (dynamic thinking for Gemini 2.5), responses may include thinking content
+	require.Contains(t, out, "--- Agent: root ---")
+	// The response should contain the answer "4" somewhere
+	require.Contains(t, out, "4")
 }
 
 func TestExec_Gemini_ToolCall(t *testing.T) {
 	out := cagentExec(t, "testdata/fs_tools.yaml", "--model=google/gemini-2.5-flash", "How many files in testdata/working_dir? Only output the number.")
 
-	require.Equal(t, "\n--- Agent: root ---\n\nCalling list_directory(path: \"testdata/working_dir\")\n\nlist_directory response → \"FILE README.me\\n\"\n1", out)
+	// With thinking enabled by default (dynamic thinking for Gemini 2.5), responses include thinking content
+	require.Contains(t, out, "--- Agent: root ---")
+	require.Contains(t, out, `Calling list_directory(path: "testdata/working_dir")`)
+	require.Contains(t, out, `list_directory response → "FILE README.me\n"`)
+	// The response should end with "1" (the count)
+	require.True(t, out != "" && out[len(out)-1] == '1', "response should end with '1'")
 }
 
 func TestExec_Mistral(t *testing.T) {

e2e/cagent_mcp_test.go

@@ -54,5 +54,6 @@ func TestMCP_MultiAgent(t *testing.T) {
 	})
 
 	require.NoError(t, err)
-	assert.Equal(t, "Hello, how can I help you today?", output.Response)
+	// Model response to "say hello" can vary, just check it contains a greeting
+	assert.Contains(t, output.Response, "Hello")
 }

e2e/testdata/cassettes/TestA2AServer_MultiAgent.yaml

@@ -8,7 +8,7 @@ interactions:
         proto_minor: 1
         content_length: 0
         host: api.openai.com
-        body: '{"messages":[{"content":"You are a multi-agent system, make sure to answer the user query in the most helpful way possible. You have access to these sub-agents:\nName: web | Description: \n\nIMPORTANT: You can ONLY transfer tasks to the agents listed above using their ID. The valid agent names are: web. You MUST NOT attempt to transfer to any other agent IDs - doing so will cause system errors.\n\nIf you are the best to answer the question according to your description, you can answer it.\n\nIf another agent is better for answering the question according to its description, call `transfer_task` function to transfer the question to that agent using the agent''s ID. When transferring, do not generate any text other than the function call.\n\n","role":"system"},{"content":"You are a knowledgeable assistant that helps users with various tasks.\nBe helpful, accurate, and concise in your responses.\n","role":"system"},{"content":"Say hello.","role":"user"}],"model":"gpt-5-mini","stream_options":{"include_usage":true},"tools":[{"function":{"name":"transfer_task","description":"Use this function to transfer a task to the selected team member.\n            You must provide a clear and concise description of the task the member should achieve AND the expected output.","parameters":{"additionalProperties":false,"properties":{"agent":{"description":"The name of the agent to transfer the task to.","type":"string"},"expected_output":{"description":"The expected output from the member (optional).","type":"string"},"task":{"description":"A clear and concise description of the task the member should achieve.","type":"string"}},"required":["agent","expected_output","task"],"type":"object"}},"type":"function"}],"stream":true}'
+        body: '{"messages":[{"content":"You are a multi-agent system, make sure to answer the user query in the most helpful way possible. You have access to these sub-agents:\nName: web | Description: \n\nIMPORTANT: You can ONLY transfer tasks to the agents listed above using their ID. The valid agent names are: web. You MUST NOT attempt to transfer to any other agent IDs - doing so will cause system errors.\n\nIf you are the best to answer the question according to your description, you can answer it.\n\nIf another agent is better for answering the question according to its description, call `transfer_task` function to transfer the question to that agent using the agent''s ID. When transferring, do not generate any text other than the function call.\n\n","role":"system"},{"content":"You are a knowledgeable assistant that helps users with various tasks.\nBe helpful, accurate, and concise in your responses.\n","role":"system"},{"content":"Say hello.","role":"user"}],"model":"gpt-5-mini","reasoning_effort":"medium","stream_options":{"include_usage":true},"tools":[{"function":{"name":"transfer_task","description":"Use this function to transfer a task to the selected team member.\n            You must provide a clear and concise description of the task the member should achieve AND the expected output.","parameters":{"additionalProperties":false,"properties":{"agent":{"description":"The name of the agent to transfer the task to.","type":"string"},"expected_output":{"description":"The expected output from the member (optional).","type":"string"},"task":{"description":"A clear and concise description of the task the member should achieve.","type":"string"}},"required":["agent","expected_output","task"],"type":"object"}},"type":"function"}],"stream":true}'
         url: https://api.openai.com/v1/chat/completions
         method: POST
       response:
@@ -17,33 +17,19 @@ interactions:
         proto_minor: 0
         content_length: -1
         body: |+
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"role":"assistant","content":"","refusal":null},"finish_reason":null}],"usage":null,"obfuscation":"c186g"}
+            data: {"id":"chatcmpl-CykKataoyjM1Yk8bJpDvOBdhXLMtu","object":"chat.completion.chunk","created":1768593812,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"role":"assistant","content":"","refusal":null},"finish_reason":null}],"usage":null,"obfuscation":"zXPgf"}
 
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}],"usage":null,"obfuscation":"57"}
+            data: {"id":"chatcmpl-CykKataoyjM1Yk8bJpDvOBdhXLMtu","object":"chat.completion.chunk","created":1768593812,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}],"usage":null,"obfuscation":"IJ"}
 
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}],"usage":null,"obfuscation":"XDOJHR"}
+            data: {"id":"chatcmpl-CykKataoyjM1Yk8bJpDvOBdhXLMtu","object":"chat.completion.chunk","created":1768593812,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":"!"},"finish_reason":null}],"usage":null,"obfuscation":"C0Oy9w"}
 
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":" How"},"finish_reason":null}],"usage":null,"obfuscation":"i1G"}
+            data: {"id":"chatcmpl-CykKataoyjM1Yk8bJpDvOBdhXLMtu","object":"chat.completion.chunk","created":1768593812,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":null,"obfuscation":"d"}
 
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":" can"},"finish_reason":null}],"usage":null,"obfuscation":"yFM"}
-
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":" I"},"finish_reason":null}],"usage":null,"obfuscation":"L8FUi"}
-
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":" help"},"finish_reason":null}],"usage":null,"obfuscation":"b0"}
-
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":" you"},"finish_reason":null}],"usage":null,"obfuscation":"UmQ"}
-
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":" today"},"finish_reason":null}],"usage":null,"obfuscation":"7"}
-
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{"content":"?"},"finish_reason":null}],"usage":null,"obfuscation":"36Ullb"}
-
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[{"index":0,"delta":{},"finish_reason":"stop"}],"usage":null,"obfuscation":"2"}
-
-            data: {"id":"chatcmpl-Cyg4WkSjCDUBxPviMs09eFjYBD3x1","object":"chat.completion.chunk","created":1768577440,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[],"usage":{"prompt_tokens":373,"completion_tokens":18,"total_tokens":391,"prompt_tokens_details":{"cached_tokens":0,"audio_tokens":0},"completion_tokens_details":{"reasoning_tokens":0,"audio_tokens":0,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"obfuscation":"NkWx"}
+            data: {"id":"chatcmpl-CykKataoyjM1Yk8bJpDvOBdhXLMtu","object":"chat.completion.chunk","created":1768593812,"model":"gpt-5-mini-2025-08-07","service_tier":"default","system_fingerprint":null,"choices":[],"usage":{"prompt_tokens":373,"completion_tokens":75,"total_tokens":448,"prompt_tokens_details":{"cached_tokens":0,"audio_tokens":0},"completion_tokens_details":{"reasoning_tokens":64,"audio_tokens":0,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"obfuscation":"CRG"}
 
             data: [DONE]
 
         headers: {}
         status: 200 OK
         code: 200
-        duration: 2.448987709s
+        duration: 2.228273792s