Skip to content

Warp token tracking #10000

@reubenbijl

Description

@reubenbijl

Pre-submit Checks

Describe the solution you'd like?

Feature Request: Automatic AI Token Tracking by Git Project and Branch

Problem

Teams using Warp’s AI features have no visibility into which projects are consuming tokens. There’s no way to answer basic questions like:

  • Which repos are driving our AI spend?
  • Is usage concentrated in a few engineers or spread across the team?
  • How much spend is happening outside of any tracked project?

Existing solutions (like AWS billing groups) require manual setup and labeling — which means they’re always incomplete and never adopted consistently.

Why manual labeling is the wrong approach

Asking teams to manually tag sessions or create project groups creates friction that guarantees poor data. Warp already knows exactly where you are: it detects the working directory and git context on every session. That information should drive attribution automatically, with zero configuration required.

If a session has no git context, that is itself a signal worth surfacing — not something to silently bury in an “untracked” bucket. A team spending significant tokens outside of any git project is a problem a team admin should be able to see.

Desired behavior

Automatic attribution — no setup required:

  • Every AI request is tagged at call time with the git remote URL and current branch
  • Attribution works from day one with no admin configuration
  • Usage rolls up to the repo level in the team dashboard, with branch as a drillable dimension

Unattributed usage is a first-class state:

  • Sessions outside a git repo appear explicitly as “Unattributed” in the team dashboard
  • Admins can see the absolute spend and percentage of total that is unattributed
  • Optionally: prompt the user inline — “You’re using AI outside a project — tag this session?”

Team dashboard surfaces:

  • Total token usage by repo (this period vs last period)
  • Drilldown to branch and user within each repo
  • Unattributed usage called out separately, not hidden
  • Export for finance/reporting purposes

What this is NOT

  • Not a manual labeling system
  • Not a billing group configuration UI that admins have to maintain
  • Not a replacement for user-level usage data — this is additive, at the project level

Prior art / inspiration

  • AWS Cost Allocation Tags (automatic tagging by resource metadata)
  • Datadog’s APM service attribution (automatic, derived from runtime context)
  • GitHub’s per-repo Actions minutes billing

Who experiences this

Engineering leads and platform/infrastructure teams at companies using Warp at scale who need to understand and control AI spend without adding process overhead to their developers.

Is your feature request related to a problem? Please describe.

No response

Additional context

No response

Operating system (OS)

Select an OS

How important is this feature to you?

4

Warp Internal (ignore) - linear-label:39cc6478-1249-4ee7-950b-c428edfeecd1

None

Metadata

Metadata

Assignees

No one assigned

    Labels

    area:agentAgent workflows, conversations, prompts, cloud mode, and AI-specific UI.area:billingPricing, plans, subscriptions, payment, and billing management.area:workspaceFile tree, workspace navigation, project switching, and working-directory management.enhancementNew feature or request.repro:highThe report includes enough evidence that the issue appears highly reproducible.triagedIssue has received an initial automated triage pass.

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions