refactor(observability): simplify run outcome tracking, route span attributes and proxy log deprecations

[kwakayama]

Follow-up cleanup of the observability code merged in #4600, #4601, #4602, #4603 and #4604. No behavior change: span attribute names, metric and label names, and log field names are unchanged.

Simplifications

• run-stream (fix(agent): mark failed agent.run spans as ERROR with their run error code #4600): the run outcome counters and the observeRunOutcomeEvent closure move out of createRuntimeAgentStreamResponse into a small createRunOutcomeTracker. The three copies of the agent.run.tool_error_count / agent.run.child_run_error_count pair (finalized, cancelled and failed paths) now come from one spanAttributes(), and the finalize log fields from logFields(). The event checks are an else if chain, since an event has only one type.
• route registry span attributes (feat(observability): tag project request spans with the serving release #4602): buildRouteRegistrySpanAttributes returns early when there is no project instead of testing projectSlug || projectId twice, which removes one nesting level from the branch/release block.
• proxy log deprecations (fix(proxy): stamp snake_case project and request fields on proxy logs #4601): the camelCase aliases on LogEntry now carry a removal condition (remove each once no Grafana dashboard, alert rule or saved Loki query filters on it) instead of a repeated "kept for Grafana dashboard transition" note. The stale "request context fields" comment is gone.
• metrics (fix(metrics): make sparse project counters countable and bound tenant series #4604): in isBoundedTenantSample, projectAttributes counted the labels that are not platform project labels. It is renamed to ownAttributes.

Reviewed and left as is

• markSpanFailed (otlp-setup), ServiceTracerSpan.markFailed and telemetryErrorType do not overlap. The first two set an error status on two different span types. telemetryErrorType classifies a thrown error. HostedAgentRunSpan.markFailed stays optional because hosts implement that public interface.
• __getTenantSeriesScopeCountForTests asserts that series registries do not outlive their targets. That memory bound has no other observable signal, and it follows the existing __getDirectTargetCountForTests pattern.

Part of veryfront/veryfront-issue-inbox#1860

Summary by CodeRabbit

• Bug Fixes
  • Run outcomes now report tool-call failures and error codes consistently across completed, cancelled, and failed runs.
  • Tenant attribute limits now apply only to tenant-owned attributes; reserved project labels do not count toward the limit. This provides more accurate tenant attribute measurements.
• Documentation
  • Clarified that camelCase log fields are deprecated aliases of their snake_case counterparts.

[greptile-apps]

Your trial has ended. Reactivate Greptile to resume code reviews.

[chatgpt-codex-connector]

You have reached your Codex usage limits for security reviews. Please try again later.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-26T23:55:37.304317Z	62ab7db	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[github-actions]

📦 Client bundle boundary

Entrypoint	Modules	Source size	Server leaks
src/index.client.ts	289	2321 KiB	✅ 0

A server module in a client graph aborts hydration in the browser. New leaks fail CI; known leaks are tracked in scripts/lint/client-bundle-baseline.json to burn down.

[coderabbitai]

Navigate logical layers of code changes, visualize relationships, and explore their blast radius.

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Advanced

Run ID: 0f4e2670-7340-4c17-87ff-29b44823f630

📥 Commits

Reviewing files that changed from the base of the PR and between 927fb54 and 62ab7db.

📒 Files selected for processing (4)

• src/internal-agents/run-stream.ts
• src/metrics/index.ts
• src/proxy/logger.ts
• src/routing/registry/registry.ts

Included review availability: This review used your included allowance. Your plan provides up to 1 included review per hour; 0 remain after this review.

---

📝 Walkthrough

Walkthrough

The changes update run outcome tracking and telemetry attribute handling. They also revise logger field deprecation comments and simplify a condition in route registry span attribute construction.

Changes

Run outcome tracking

Layer / File(s)	Summary
Track outcomes through terminal reporting src/internal-agents/run-stream.ts	A tracker collects tool-call errors, errors from child-agent tool calls, and the first stable RunError code. Terminal error selection, span attributes, and log fields use the tracker. The AgentRunTerminalError fallback remains in place.

Tenant attribute sampling

Layer / File(s)	Summary
Count tenant-owned attributes src/metrics/index.ts	The tenant sample count now tracks tenant-owned attributes. Reserved project labels remain excluded from the limit.

Logger field documentation

Layer / File(s)	Summary
Document field aliases src/proxy/logger.ts	Comments mark camelCase fields as deprecated aliases and direct users to the corresponding snake_case fields.

Route registry span attributes

Layer / File(s)	Summary
Gate route attribute construction src/routing/registry/registry.ts	The function returns the existing attributes when both projectSlug and projectId are absent. Attribute construction is unchanged when either is present.

Priority: ⬇️ Low

Estimated code review effort: 3 (Moderate) | ~20 minutes

Change: Refactor

Merge Risk: ⚪ Minimal · up to 62ab7

No actionable issue remains in the supplied review evidence; the PR is mergeable after normal checks.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 33.33% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 6 functions across 4 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main observability refactoring, including run outcome tracking, route span attributes, and proxy log deprecations.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

• Fix all pre-merge checks with AI

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Commit to this branch
• Create a new PR

🧪 Generate unit tests (beta)

• Commit to this branch
• Create a new PR

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[gitar-bot]

Important

You are using the Gitar free plan. Upgrade to unlock code review, CI analysis, auto-apply, custom automations, and more.

_Gitar

[kwakayama]

Code review: 90/100

Clean, low-risk refactor. Traced each change against its pre-refactor equivalent; behavior is unchanged.

• createRunOutcomeTracker in run-stream.ts consolidates three duplicated spanAttributes/logFields blocks (finalized, cancelled, failed) into one factory. The else if chain is safe because event only ever matches one of ToolCallStart, ToolCallResult, or RunError per call.
• buildRouteRegistrySpanAttributes's early return (if (!projectSlug && !projectId) return attributes;) is logically equivalent to the original (projectSlug || projectId) guard on both blocks, and removes a nesting level as described.
• ownAttributes rename and the LogEntry doc-comment cleanup in proxy/logger.ts are cosmetic only, no logic touched.
• PR description is precise: it maps each simplification to the PR that introduced it and calls out what was reviewed and deliberately left alone (markSpanFailed vs. ServiceTracerSpan.markFailed vs. telemetryErrorType, the tenant-scope test helper). That's a good practice for a cleanup PR.
• No test changes accompany this, which is fine for a pure refactor with existing coverage, but I could not execute deno task test:file in this environment (no deno binary available) to confirm run-stream.test.ts and registry.test.ts still pass on this exact head. CI's test/coverage shards were still in progress at review time, worth confirming they land green before merge.

Minor nit, non-blocking: in createRunOutcomeTracker, toolCallId is destructured from payload but toolCallName is accessed as payload.toolCallName in the same condition. Consistent either way would read slightly cleaner, but it's not a defect.

---

Generated by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud