Cost per session for Claude and Pi, like Codex has #4930

Activity

1. clawsweeper

   added

   clawsweeper:intake:featurePreliminary submission classification; full source review pending.Preliminary submission classification; full source review pending.

   P2Normal priority bug or improvement with limited blast radius.Normal priority bug or improvement with limited blast radius.

   clawsweeper:needs-maintainer-reviewClawSweeper marked this issue as needing maintainer review before automation.ClawSweeper marked this issue as needing maintainer review before automation.

   clawsweeper:needs-product-decisionClawSweeper marked this issue as needing a product or behavior decision.ClawSweeper marked this issue as needing a product or behavior decision.

   clawsweeper:no-new-fix-prClawSweeper does not recommend queueing a new automated fix PR for this issue.ClawSweeper does not recommend queueing a new automated fix PR for this issue.

   clawsweeper:source-reproClawSweeper found a high-confidence source-level issue reproduction.ClawSweeper found a high-confidence source-level issue reproduction.

   impact:ux-frictionUser-facing flow adds avoidable confusion or support burden without fully blocking progress.User-facing flow adds avoidable confusion or support burden without fully blocking progress.

   issue-rating: 🦞 diamond lobsterVery strong issue quality with high-confidence source-level or clear reproduction.Very strong issue quality with high-confidence source-level or clear reproduction.

   and removed

   clawsweeper:intake:featurePreliminary submission classification; full source review pending.Preliminary submission classification; full source review pending.

   on Nov 4, 2026
2. clawsweeper commented on Nov 4, 2026

   

   clawsweeperBot

   on Nov 4, 2026 – with ClawSweeper

   Last edited by clawsweeper

   More actions

   Codex review: this still needs some work.

   Summary
   Keep open: Claude and Pi still lack session/project cost breakdowns on main, and the repository requires conservative handling of feature requests.

   Reproducibility: Current source and explicit CLI tests establish the reported omission; no command against real accounts or live reproduction was run.

   Next step
   Decide whether to prioritize Claude/Pi session and project parity, with task-prefix aggregation considered separately.

   Review details

   Best possible solution:

   Extend the existing breakdown models with Claude/Pi adapters and shared presentation while preserving privacy, accounting, and estimate labels.

   Do we have a high-confidence way to reproduce the issue?

   Current source and explicit CLI tests establish the reported omission; no command against real accounts or live reproduction was run.

   Is this the best way to solve the issue?

   Provider adapters feeding the existing breakdown models are appropriate; displaying recorded Pi costs or merging task-name prefixes would require additional policy beyond presentation parity.

   AGENTS.md: found and applied where relevant.

   Remaining risk / open question:

   • Session attribution must preserve existing duplicate suppression, provider source ownership, partial-history coverage, and pricing semantics.
   • Task-prefix aggregation has no defined identity or collision rules and needs a separate scope decision.

   Codex review notes: model internal, reasoning medium; reviewed against 897bff31731b.

   Label changes

   Label changes:

   No label changes.

   Label justifications:

   • P2: Useful parity improvement with limited blast radius; aggregate cost reporting remains available.
   • impact:ux-friction: Users cannot inspect Claude/Pi task costs through the conversation views and grouping commands already available for Codex.

   Evidence reviewed

   Acceptance criteria:

   • swift test --filter CLICostTests
   • swift test --filter CostUsageFetcherTests
   • swift test --filter PiSessionCostScannerTests
   • swift test --filter CostHistoryChartMenuViewTests
   • swift test --filter SpendDashboardSessionRowTests

   What I checked:

   • Grouping is deliberately Codex-only: Provider filtering excludes Claude and Pi from grouped text reports, and the warning matches the report. (Sources/CodexBarCLI/CLICostCommand.swift:603, 897bff31731b)
   • Both requested adapters remain missing: Pi explicitly publishes empty projects and sessions; the shared native scan populates those breakdowns only for Codex. (Sources/CodexBarCore/CostUsageFetcher.swift:733, 897bff31731b)
   • Existing tests define the current boundary: The session-grouping test explicitly expects text output to retain Codex alone while JSON retains other providers. (Tests/CodexBarTests/CLICostTests.swift:83, 897bff31731b)
   • Pi titles already have a parsing seam: The local session metadata parser reads session_info.name; that capability is separate from producing conversation cost breakdowns. (Sources/CodexBarCore/PiFamilySessionScanner.swift:102, 897bff31731b)
   • Preserve established pricing semantics: Pi documentation specifies independent API-price estimates rather than substituting recorded usage.cost, and describes source ownership that avoids counting Pi usage twice. (docs/pi.md:16, 897bff31731b)
   • Prior work provides foundations: feat(cli): add Codex session cost grouping #2854 implemented native Codex CLI grouping; feat(cost): show conversation titles in Cost menu #4434 added existing Codex titles to the menu. Its downloaded synthetic screenshots demonstrate title and privacy rendering, not Claude/Pi support.

   How this review workflow works

   • ClawSweeper keeps one durable marker-backed review comment per issue or PR.
   • Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
   • A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
   • PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
   • Maintainers can also comment @clawsweeper review to request a fresh review only.
   • Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge; those execution lanes are retired.
   • Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

   Reviewed November 4, 2026, 3:34 PM ET / 20:34 UTC.
3. steipete commented on Nov 7, 2026

   

   steipete

   on Nov 7, 2026

   Owner

   More actions

   Thanks for describing the log formats. We checked the current command: session/project grouping and the Conversations view remain Codex-only. Adding Claude and Pi needs persistent session/project attribution, deduplication, incremental updates, and matching UI behavior; it is not just enabling the existing CLI switch. We're closing this parity request rather than keeping it as an open commitment. A focused implementation for one provider with synthetic ownership and append/update coverage would be a basis to reconsider. Existing ungrouped cost totals remain available.
4. steipete

   closed this as not plannedon Nov 7, 2026
5. oryband commented on Nov 8, 2026

   

   oryband

   on Nov 8, 2026

   ContributorAuthor

   More actions

   I opened two focused PRs for this, one per provider, as you suggested:

   • Pi: Pi: cost per session and per project #5025
   • Claude: Claude: cost per session and per project #5026

   Each one stores session and project attribution in the cache, keeps the existing dedup, handles appends, and adds --group-by session|project plus the menu views. All tests use synthetic data.