Include cached tokens in copilot debug log file output

[petercort]

PR implements #311186.

Simply includes the USAGE_CACHE_READ_INPUT_TOKENS as pulled from OTEL. updated the writes LLM request with token counts test with a cachedTokens value.

Testing:

1. Run unit tests
2. Build and run the code-oss-dev

• Open a directory
• Spawn a GitHub Copilot chat interaction that will hit the LLM
• Load the debug log, on my macos setup ~/Library/Application Support/code-oss-dev/User/workspaceStorage/<workspace_id>/GitHub.copilot-chat/debug-logs/<session_id>/main.jsonl
• Search for cachedTokens
  • New output in my testing looks like {"model":"claude-opus-4.6","inputTokens":22891,"outputTokens":229,"cachedTokens":22559,

[Copilot]

Pull request overview

This PR implements #311186 by extending Copilot Chat’s debug file logging (JSONL) to include cached input token counts derived from OTel spans, aligning file output more closely with the debug UI’s token breakdown.

Changes:

• Add cachedTokens to llm_request debug log entries when GenAiAttr.USAGE_CACHE_READ_INPUT_TOKENS is present.
• Update the unit test to emit the cache-read token attribute and assert it is written to the log.

Show a summary per file

File	Description
extensions/copilot/src/extension/chat/vscode-node/chatDebugFileLoggerService.ts	Emits attrs.cachedTokens in JSONL output for chat spans when the OTel cached-token attribute is present.
extensions/copilot/src/extension/chat/vscode-node/test/chatDebugFileLoggerService.spec.ts	Updates the chat span test helper and asserts cachedTokens is logged.

Copilot's findings

• Files reviewed: 2/2 changed files
• Comments generated: 0

[petercort]

@microsoft-github-policy-service agree company="GitHub"