feat(tools): compress tool schemas and truncate descriptions

[juslintek]

What

Adds opt-in tool schema compression and description truncation to reduce token usage when many MCP tools are present.

Why

Claude Code with MCP servers can send 100+ tools per request. Each tool's JSON Schema includes verbose fields (description, title, examples, $comment, markdownDescription) that consume context window tokens but aren't needed for tool call generation. With 134 tools observed in live traffic, this saves thousands of tokens per request.

How

compress_json_schema() recursively strips verbose fields from JSON Schema properties, keeping only structural fields (type, properties, required, enum, items, etc.).

Long tool descriptions are also truncated to a configurable limit.

Changes

• config.py: Add KIRO_COMPRESS_TOOL_SCHEMAS (default: true) and KIRO_TOOL_DESC_MAX_CHARS (default: 200)
• converters_core.py: Add compress_json_schema() function, integrate into convert_tools_to_kiro_format()

Config

# Enable/disable schema compression (default: true)
KIRO_COMPRESS_TOOL_SCHEMAS=true

# Max description length, 0 to disable (default: 200)
KIRO_TOOL_DESC_MAX_CHARS=200

Testing

All 1413 existing tests pass. No dependencies on other PRs — this is independent.

[cla-bot]

Thanks for the PR! 🎉

Before merge, we need a one-time CLA confirmation.
It confirms that you have the right to contribute this code and allow the project to use it.

Full CLA text:
https://github.com/jwadow/kiro-gateway/blob/main/CLA.md

Please reply once with:

I have read the CLA and I accept its terms

You need to write once, all further messages from me can be ignored.

[juslintek]

I have read the CLA and I accept its terms

[kosuke-zhang]

I have read the CLA and I accept its terms

Author has been inactive for a long time... Could you fork this repo and maintain it?!