Skip to content

Monitor non-interactive Python environment tooling after release #26081

Description

@eleanorjboyd

Background

#25940 will change environment-tool behavior so auto-approved agent calls do not block on extension-owned UI. Because this is a user-facing behavior change in environment creation and selection, we need release monitoring that can detect regressions before broadening the contract or transferring ownership under #26078.

Define privacy-conscious measures

Where existing telemetry policy permits, measure aggregate outcomes such as:

  • invocation and completion counts;
  • success, structured failure, cancellation, and timeout rates;
  • whether an agent invocation requested user-input UI;
  • environment reuse versus creation;
  • creation failure category;
  • retries or repeated calls within a task/session;
  • time to completion;
  • active backend/capability tier;
  • Python Environments enabled, disabled, or unavailable state.

Do not collect workspace paths, environment paths, package names, dependency contents, prompts, or other sensitive workspace details.

Rollout plan

  • Establish a pre-change baseline where comparable data exists.
  • Define dashboards or queries for the first pre-release and stable releases containing the fix.
  • Compare completion, failure, cancellation, latency, retry, and unexpected-interaction rates before and after release.
  • Review qualitative issue and feedback trends alongside telemetry.
  • Define thresholds for pausing rollout, reverting behavior, or opening follow-up bugs.
  • Record limitations where the platform cannot distinguish tool approval from extension-owned UI.

Acceptance criteria

Metadata

Metadata

Assignees

Labels

area-environmentsFeatures relating to handling interpreter environments

Type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions