What metrics would you like to measure from Gemini CLI?

[blanca-delgado-parra]

Hello Gemini CLI community,
We are actively thinking about Open Telemetry, Metrics and Out of the box dashboards to help you understand your/your team’s usage of Gemini CLI. We would love to hear from you on what metrics you would be interested in. Feel free leave a comment with your thoughts!

Thank you!

[kev-pebble]

Accepted suggestions, Edits.

[arturroo]

Used token count and/or for free tier how much tokens left.

[arturroo]

Used in/out tokens per prompt/answer or per chat or per day or during last 24h

[cartermp]

• Tokens by:
  • Tool call
  • Session
  • Model
• Error types and frequency

And TBH just what Claude Code captures too: https://docs.anthropic.com/en/docs/claude-code/monitoring-usage#metrics

[TayDa64]

I would like to see plaintext; websites Gemini is referencing/grounding itself with. I can build the markdown instructions, but without confirmation, it is difficult to narrow down on the misunderstandings.

[arturroo]

Current memory usage so we know what how each prompt influence the stability of the tool so we know what to avoid in order for the tool not to break.

[palladius]

1. Tool call / mistakes count
2. Context length
3. Cross-correlate the above, mjaybe with bucketing (mistakes_count at 0 context vs 100k context). 10-20 buckets should suffice.