feat/claude 1m context 20250822172021

[jacsamell]

User description

• feat: auto-set Anthropic 1M-context beta header for Claude models; merge with user extra_headers; OpenAI type alias shim for LiteLLM
• feat: switch defaults to claude-sonnet-4-20250514; add aliases and MAX_TOKENS entry; update docs and action inputs

---

PR Type

Enhancement

---

Description

• Automatically set Anthropic beta header for 1M context
• Switch default model to claude-sonnet-4-20250514
• Enhance header merging for LiteLLM integration
• Update model token limits and configurations

---

Changes walkthrough 📝

	Relevant files
Configuration changes	__init__.py Update model configuration to Sonnet 4                                      pr_agent/algo/init.py • Replace claude-3-7-sonnet-20250219 with claude-sonnet-4-20250514 • Update token limit from 200k to 1M context • Fix trailing whitespace in model list +2/-2
Enhancement	litellm_ai_handler.py Enhanced Claude integration with 1M context                            pr_agent/algo/ai_handlers/litellm_ai_handler.py • Auto-set anthropic-beta header for Claude 1M context • Merge user extra_headers with existing headers • Add error handling for header configuration • Exclude Bedrock models from beta header logic +17/-2

---

Need help?

Type /help how to ... in the comments thread for any questions about PR-Agent usage.

Check out the documentation for more information.

[github-actions]

PR Reviewer Guide 🔍

(Review updated until commit e948cce)

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 3 🔵🔵🔵⚪⚪

AI Review Score: 76%

🧪 No relevant tests

Confidence score [1-100]: 85

Complexity score [1-10]: 5

Security score [1-10]: 7

Auto approve recommendation: false

Auto approve reasoning: The PR has a duplicate model entry that could cause issues, uses broad exception handling, and lacks tests for the new functionality. While the header merging logic is sound, these concerns prevent auto-approval.

Requires human approval: Code Quality Issues

💡 Code suggestions Improve Model Detection Use a more robust method for detecting Anthropic Claude models instead of string matching which could be error-prone. Existing code: if ( ("claude" in model.lower() or model.lower().startswith("anthropic/")) and not model.lower().startswith("bedrock/") ): Improved code: # Define Claude model prefixes for better maintainability claude_prefixes = ["anthropic/claude", "claude-"] bedrock_prefixes = ["bedrock/"] model_lower = model.lower() is_claude = any(model_lower.startswith(prefix) for prefix in claude_prefixes) is_bedrock = any(model_lower.startswith(prefix) for prefix in bedrock_prefixes) if is_claude and not is_bedrock: Specific Exception Handling Use more specific exception handling to better identify and log potential issues with header setting. Existing code: except Exception as e: get_logger().debug(f"Unable to set Anthropic beta header: {e}") Improved code: except (KeyError, TypeError, ValueError) as e: get_logger().debug(f"Unable to set Anthropic beta header: {e}") except Exception as e: get_logger().warning(f"Unexpected error setting Anthropic beta header: {e}")

🔒 No security concerns identified

⚡ Recommended focus areas for review Duplicate Entry The model anthropic/claude-sonnet-4-20250514 is being added to MAX_TOKENS but it already exists on line 78. This creates a duplicate key in the dictionary which may cause unexpected behavior. 'anthropic/claude-sonnet-4-20250514': 1000000, Broad Exception The exception handling for setting Anthropic beta header catches all exceptions with a generic Exception. This could mask important errors and make debugging difficult. except Exception as e: get_logger().debug(f"Unable to set Anthropic beta header: {e}")

---

🔍 Manual review required

Reason: Manual review required: Code Quality Issues