[voltcrash]

The macOS resource monitor retained command lines and memory fields for unrelated processes in its long-lived discovery table. Reuse Linux's existing two-stage collection on macOS: keep discovery lean, then load detailed data for tracked processes and the monitor process needed for start-time calibration. Windows keeps its existing path.

Continues voltcrash's work in #19745; contributor commits are preserved and main was merged normally. Fresh independent reconstruction matched the original implementation exactly, so no implementation churn or dependency change was added.

Verification on eed0c76:

• Fresh Astra test pass retained the single added real-collector regression: 18 tests before→18 after (main has 17). Linux cargo test --locked, formatting, and Clippy with warnings denied pass.
• Linux production-revert spotcheck also passes because Linux already used two-stage collection. This is not macOS red/green proof.
• Actual Linux binary JSON protocol verified an owned process being added then removed: snapshot/history process counts 0→1→0; command, resident memory, and start time present while tracked; process table includes that process. Clean shutdown, protocol v3.
• Source inspection of locked sysinfo 0.39.3 confirms reduced retained command/memory fields. macOS still enumerates all PIDs and reads argument buffers for names and CPU task information; this does not promise elimination of all unrelated OS reads or a measured speedup.

Verification output

Target-platform limit: this Linux host did not compile or execute macOS, and the author's macOS measurements were not independently reproduced. macOS runtime verification and maintainer approval remain for a human. Windows, Electron desktop, iOS and Android were not exercised. No visual UI change; screenshots are not applicable.

One independent full-diff review round from Opus 5.5 and Sol 6 found no code blockers. Their description-accuracy finding is reflected above. Unmerged; latest-head CI and maintainer approval remain required.

Models/harnesses: tests Codex gpt-6-astra; implementation Codex gpt-6.1-sol through T3 Code; reviews Claude claude-opus-5-5 and Codex gpt-6-sol.

[macroscopeapp]

Approvability

Verdict: Needs Human Approval

Macroscope's review found this PR not approvable — The change alters the existing macOS sampling path to automatically suppress memory, disk, and command-detail collection for untracked processes, then refresh those details only for tracked PIDs. The implementation is isolated and tested, but this is a material runtime processing gate that merits human assessment.

^{You can add or adjust custom eligibility rules. Learn more.}

[coderabbitai]

Warning

Review limit reached

Only developers with an assigned seat can use this organization's usage-based review budget, and seats here are assigned manually. Ask an admin to assign a seat, or change the review continuation mode in Billing.

Next included review available in 10 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used all 10 included reviews currently available.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

• Configuration used: Path: .coderabbit.config.ts
• Review profile: CHILL
• Plan: Advanced
• Run ID: 5052fa5b-fce7-4368-b874-aa0c0c591d98

📥 Commits

Reviewing files that changed from the base of the PR and between 49eb1d2 and 6936496.

📒 Files selected for processing: 1

• native/resource-monitor/src/main.rs

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

• Configuration used: Path: .coderabbit.config.ts
• Review profile: CHILL
• Plan: Advanced
• Run ID: 8dda7578-216c-493a-b958-082b3e1541b3

📥 Commits

Reviewing files that changed from the base of the PR and between e536bfd and 49eb1d2.

📒 Files selected for processing: 1

• native/resource-monitor/src/main.rs

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 9 remain after this review.

📝 Walkthrough

Linux and macOS now use CPU-only process discovery followed by a separate detail refresh for selected processes. Other platforms retain full refresh during discovery. A platform-gated test checks the process data returned by these steps.

Changes: Two-stage process refresh

Layer / File(s)	Summary
Platform-gated process refresh native/resource-monitor/src/main.rs	A compile-time flag enables two-stage refresh on Linux and macOS. Discovery uses CPU-only refresh, and tracked processes receive a separate detail refresh. A test checks details for selected processes and confirms that unselected processes lack command and memory data.

Priority: ⬇️ Low

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Bug fix

Suggested reviewers: t3dotgg

Merge Risk: ⚪ Minimal · up to 49eb1

No actionable merge-blocking risk is identified for the macOS process-sampling change.

Security architecture risk: ⚪ Minimal · up to 49eb1

The change reduces retention of unrelated processes’ command data without adding an interface or expanding process-selection authority. Existing identity checks remain in the detail-output path. No material security risk introduced by this change was identified.

Retained concerns
No architecture-level concerns identified.

Security review details

Security Blast Radius

• inferred — The changed exposure is local macOS process-detail collection and retention. The examined diff does not expand the existing stdin caller’s ability to configure roots or external PIDs, nor add a remote interface or privilege-granting operation.

Trust Boundaries and Controls

• observed — Configured external identities are checked against discovery start times when an expected time is supplied. Descendant traversal rejects children older than their observed parent, and detail output requires calibrated start-time agreement. Root selection remains PID-based; these predicates predate this PR and use the existing timestamp precision.

Resilience and Maintainability Implications

• observed — Sampling and configuration mutations are serialized in the event loop. Selection changes and shutdown wait for the active sample to return, preventing a sample from observing a partially updated configuration. Each subsequent two-stage sample creates a fresh detail table.

Pre-merge checks: 4

Check name	Status	Explanation
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.
Title check		The title clearly identifies the resource-monitor fix and the macOS process-detail retention issue.
Description check		The description explains the problem, implementation, scope, verification results, platform limitations, and remaining approval requirements. It does not use the template headings, but it provides the…

Full details: Description check

Explanation

The description explains the problem, implementation, scope, verification results, platform limitations, and remaining approval requirements. It does not use the template headings, but it provides the required information in a complete narrative.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}

[juliusmarminge]

Taking over this one as maintainer for batch usage-billing-01. I will keep the macOS process-detail retention fix focused, independently check its tests and implementation, and push follow-up commits to this branch. No merge is planned.