feat: Improve tool call extraction and stripping to support multiple …

[RahulChand028]

📝 Description

Harden the tool call extraction logic by implementing a robust, state-aware scanner and supporting complex model outputs.

Key Changes

• Robust Parsing: Implemented a state-aware findMatchingBrace that correctly handles nested objects, JSON strings, and escaped characters.
• Improved Detection: Added support for markdown code block wrappers (json ... ) and flexible JSON markers (e.g., { "tool_calls": ... }).
• Multi-Block Extraction: Enhanced logic to extract and strip multiple tool call blocks from a single response.
• Refactored Utility: Centralized scanning logic into a nextToolCallBlock helper to improve maintainability and performance.
• Comprehensive Tests: Added a full test suite covering various edge cases and ensuring no regressions.

🗣️ Type of Change

• 🐞 Bug fix (non-breaking change which fixes an issue)
• ✨ New feature (non-breaking change which adds functionality)
• 📖 Documentation update
• ⚡ Code refactoring (no functional changes, no api changes)

🤖 AI Code Generation

• 🤖 Fully AI-generated (100% AI, 0% Human)
• 🛠️ Mostly AI-generated (AI draft, Human verified/modified)
• 👨‍💻 Mostly Human-written (Human lead, AI assisted or none)

🔗 Related Issue

I raised a fix earlier, but some unrelated changes were accidentally included while resolving conflicts. I’ve now created a new PR with the improvements and without those extra changes.
#497

📚 Technical Context (Skip for Docs)

• Reference URL:
• Reasoning: The previous simplistic brace counter was fragile when encountering braces in regex, grep commands, or escaped JSON strings. These improvements ensure CLI-based providers like Claude and Codex can reliably extract structured data from diverse LLM outputs.

🧪 Test Environment

• Hardware: PC
• OS: Linux (Ubuntu/Debian)
• Model/Provider: Verified via Go Unit Tests and Claude CLI mock
• Channels: CLI

📸 Evidence (Optional)

Details

☑️ Checklist

• My code/docs follow the style of this project.
• I have performed a self-review of my own changes.
• I have updated the documentation accordingly.