Add requirements.txt and update boto3 minimum to 1.42.0

TTL support in cachePoint requires boto3 >= 1.42.0 (older versions
lack the ttl field in CachePointBlock). Consolidated dependency into
a single requirements.txt and updated all README/notebook references.

Author: balak4

introduction-to-bedrock/prompt-caching/README.md

@@ -60,7 +60,7 @@ The **Converse API** `cachePoint` syntax is identical across all model families.
 
 - AWS account with Amazon Bedrock access
 - Model access enabled for the target model
-- Python 3.9+ with `boto3 >= 1.35.76`
+- Python 3.9+ with dependencies: `pip install -r requirements.txt`
 - AWS credentials configured (default profile or environment variables)
 
 ## Additional Resources

introduction-to-bedrock/prompt-caching/converse_api/notebooks/02_system_prompt_caching.ipynb

@@ -4,7 +4,7 @@
    "cell_type": "markdown",
    "id": "intro",
    "metadata": {},
-   "source": "# System Prompt Caching — Converse API\n\nThis notebook demonstrates how to cache system prompts — persona definitions, guidelines, and domain knowledge that stay constant across multiple user messages — using the **Converse** and **ConverseStream** APIs.\n\n## What you'll learn\n\n- **Converse API**: Cache system prompts using `cachePoint` in the `system` parameter\n- **ConverseStream API**: Same syntax with streaming response parsing\n- Practical pattern: same cached persona, varying user queries\n\n## Prerequisites\n\n- AWS credentials configured (default profile)\n- Access to Claude Sonnet 4.6 on Amazon Bedrock\n- `boto3 >= 1.35.76`\n\n## Token threshold\n\nClaude Sonnet 4.6 requires at least **2,048 tokens** per cache checkpoint."
+   "source": "# System Prompt Caching — Converse API\n\nThis notebook demonstrates how to cache system prompts — persona definitions, guidelines, and domain knowledge that stay constant across multiple user messages — using the **Converse** and **ConverseStream** APIs.\n\n## What you'll learn\n\n- **Converse API**: Cache system prompts using `cachePoint` in the `system` parameter\n- **ConverseStream API**: Same syntax with streaming response parsing\n- Practical pattern: same cached persona, varying user queries\n\n## Prerequisites\n\n- AWS credentials configured (default profile)\n- Access to Claude Sonnet 4.6 on Amazon Bedrock\n- `boto3 >= 1.42.0`\n\n## Token threshold\n\nClaude Sonnet 4.6 requires at least **2,048 tokens** per cache checkpoint."
   },
   {
    "cell_type": "code",

introduction-to-bedrock/prompt-caching/converse_api/notebooks/03_tool_definition_caching.ipynb

@@ -30,7 +30,7 @@
     "\n",
     "- AWS credentials configured (default profile)\n",
     "- Access to Claude Sonnet 4.6 on Amazon Bedrock\n",
-    "- `boto3 >= 1.35.76`\n",
+    "- `boto3 >= 1.42.0`\n",
     "\n",
     "## Token threshold\n",
     "\n",

introduction-to-bedrock/prompt-caching/converse_api/notebooks/04_mixed_ttl_caching.ipynb

@@ -24,7 +24,7 @@
     "\n",
     "- AWS credentials configured (default profile)\n",
     "- Access to Claude Sonnet 4.6 on Amazon Bedrock\n",
-    "- `boto3 >= 1.35.76`\n",
+    "- `boto3 >= 1.42.0`\n",
     "\n",
     "## Token threshold\n",
     "\n",

introduction-to-bedrock/prompt-caching/converse_api/notebooks/05_tenant_isolation.ipynb

@@ -28,7 +28,7 @@
     "\n",
     "- AWS credentials configured (default profile)\n",
     "- Access to Claude Sonnet 4.6 on Amazon Bedrock\n",
-    "- `boto3 >= 1.35.76`"
+    "- `boto3 >= 1.42.0`"
    ]
   },
   {

introduction-to-bedrock/prompt-caching/converse_api/notebooks/06_langchain_integration.ipynb

@@ -19,7 +19,7 @@
     "\n",
     "- AWS credentials configured (default profile)\n",
     "- Access to Claude Sonnet 4.6 on Amazon Bedrock\n",
-    "- `langchain-aws >= 0.2.12`, `boto3 >= 1.35.76`"
+    "- `langchain-aws >= 0.2.12`, `boto3 >= 1.42.0`"
    ]
   },
   {

introduction-to-bedrock/prompt-caching/converse_api/scripts/README.md

@@ -46,7 +46,7 @@ Modify these values as needed before running. The `cachePoint` syntax is model-a
 ## Requirements
 
 - Python 3.9+
-- `boto3 >= 1.35.76`
+- `pip install -r ../../requirements.txt`
 - AWS credentials configured for the specified profile
 - Access to the specified model on Amazon Bedrock
 

introduction-to-bedrock/prompt-caching/invoke_model_api/anthropic/notebooks/01_message_content_caching.ipynb

@@ -28,7 +28,7 @@
     "\n",
     "- AWS credentials configured (default profile)\n",
     "- Access to Claude Sonnet 4.6 on Amazon Bedrock\n",
-    "- `boto3 >= 1.35.76`\n",
+    "- `boto3 >= 1.42.0`\n",
     "\n",
     "## Token threshold\n",
     "\n",

introduction-to-bedrock/prompt-caching/invoke_model_api/anthropic/notebooks/02_system_prompt_caching.ipynb

@@ -4,7 +4,7 @@
    "cell_type": "markdown",
    "id": "intro",
    "metadata": {},
-   "source": "# System Prompt Caching — InvokeModel API (Anthropic)\n\nThis notebook demonstrates how to cache system prompts using the **InvokeModel** and **InvokeModelWithResponseStream** APIs with Anthropic Claude's native `cache_control` syntax.\n\n## What you'll learn\n\n- **InvokeModel API**: Cache system prompts using `cache_control` in the `system` array\n- **InvokeModelWithResponseStream**: Same syntax with streaming response parsing\n- Practical pattern: same cached persona, varying user queries\n\n## Anthropic-specific syntax\n\n```python\n\"system\": [{\n    \"type\": \"text\",\n    \"text\": \"<system prompt>\",\n    \"cache_control\": {\"type\": \"ephemeral\", \"ttl\": \"5m\"}\n}]\n```\n\nFor the model-agnostic Converse API syntax (`cachePoint`), see [converse_api/](../../converse_api/).\n\n## Prerequisites\n\n- AWS credentials configured (default profile)\n- Access to Claude Sonnet 4.6 on Amazon Bedrock\n- `boto3 >= 1.35.76`\n\n## Token threshold\n\nClaude Sonnet 4.6 requires at least **2,048 tokens** per cache checkpoint."
+   "source": "# System Prompt Caching — InvokeModel API (Anthropic)\n\nThis notebook demonstrates how to cache system prompts using the **InvokeModel** and **InvokeModelWithResponseStream** APIs with Anthropic Claude's native `cache_control` syntax.\n\n## What you'll learn\n\n- **InvokeModel API**: Cache system prompts using `cache_control` in the `system` array\n- **InvokeModelWithResponseStream**: Same syntax with streaming response parsing\n- Practical pattern: same cached persona, varying user queries\n\n## Anthropic-specific syntax\n\n```python\n\"system\": [{\n    \"type\": \"text\",\n    \"text\": \"<system prompt>\",\n    \"cache_control\": {\"type\": \"ephemeral\", \"ttl\": \"5m\"}\n}]\n```\n\nFor the model-agnostic Converse API syntax (`cachePoint`), see [converse_api/](../../converse_api/).\n\n## Prerequisites\n\n- AWS credentials configured (default profile)\n- Access to Claude Sonnet 4.6 on Amazon Bedrock\n- `boto3 >= 1.42.0`\n\n## Token threshold\n\nClaude Sonnet 4.6 requires at least **2,048 tokens** per cache checkpoint."
   },
   {
    "cell_type": "code",

introduction-to-bedrock/prompt-caching/invoke_model_api/anthropic/notebooks/03_tool_definition_caching.ipynb

@@ -30,7 +30,7 @@
     "\n",
     "- AWS credentials configured (default profile)\n",
     "- Access to Claude Sonnet 4.6 on Amazon Bedrock\n",
-    "- `boto3 >= 1.35.76`\n",
+    "- `boto3 >= 1.42.0`\n",
     "\n",
     "## Token threshold\n",
     "\n",