GITHUB-COPILOT-CLI PUB_DATE: 2026.05.12

COPILOT CLI 1.0.45: OPENTELEMETRY SPANS AND AN /AUTOPILOT SWITCH

GitHub Copilot CLI now emits standard OpenTelemetry spans for agent tool calls and adds an /autopilot mode toggle. The [v1.0.45 release](https://github.com/git...

Copilot CLI 1.0.45: OpenTelemetry spans and an /autopilot switch

GitHub Copilot CLI now emits standard OpenTelemetry spans for agent tool calls and adds an /autopilot mode toggle.

The v1.0.45 release aligns Copilot CLI telemetry with GenAI semantic conventions: MCP tool calls are reported as tool_call spans, and a new gen_ai.client.operation.duration metric times tool runs. It also adds /autopilot and /fork commands, faster startup, and a session-resume fix.

If you’ve seen Copilot agents stall or go off-script, you can now observe and rein them in. Use spans and timings to trace behavior, and keep automation opt-in with /autopilot. For teams leaning into spec-first workflows, GitHub’s new Spec Kit points to where governance is headed; the community is already flagging erratic agent behavior here.

[ WHY_IT_MATTERS ]
01.

You can finally trace Copilot agent tool calls in your existing OpenTelemetry stack with standard spans and a useful duration metric.

02.

The /autopilot toggle gives teams tighter control over when agents make changes, reducing surprise edits.

[ WHAT_TO_TEST ]
  • terminal

    Enable OTel export, run a CLI session that triggers MCP tools, and verify tool_call spans plus gen_ai.client.operation.duration show up with sane p95s.

  • terminal

    Run the same task in interactive vs /autopilot modes on a protected repo and confirm no unintended writes or privilege bypass.

[ BROWNFIELD_PERSPECTIVE ]

Legacy codebase integration strategies...

  • 01.

    Plumb Copilot CLI spans into your OpenTelemetry collector and map attributes to your existing dashboards and alerts.

  • 02.

    Default agents to interactive; restrict /autopilot via RBAC and branch protections; validate the session-resume fix with extension permission prompts.

[ GREENFIELD_PERSPECTIVE ]

Fresh architecture paradigms...

  • 01.

    Stand up basic agent SLOs using operation duration and error rates, and alert on tool stalls or loops.

  • 02.

    Pilot spec-first workflows with Spec Kit to bound agent scope and verify outputs before enabling /autopilot.

Enjoying_this_story?

Get daily GITHUB-COPILOT-CLI + SDLC updates.

  • Practical tactics you can ship tomorrow
  • Tooling, workflows, and architecture notes
  • One short email each weekday

FREE_FOREVER. TERMINATE_ANYTIME. View an example issue.

GET_DAILY_EMAIL
AI + SDLC // 5 MIN DAILY