[juliusmarminge]

makeClaudeQueryOptions sent a filled-in "Auto-compact after" value only as the autoCompactWindow key in --settings. Claude Code ranks CLAUDE_CODE_AUTO_COMPACT_WINDOW above that key, so a user who set the variable in ~/.claude/settings.json had T3's value silently ignored, while T3's meter (resolveAutoCompactWindow, which puts the T3 setting first) still drew the tick at the T3 window. The fix also sends the value as env.CLAUDE_CODE_AUTO_COMPACT_WINDOW in the launch settings' env map, merged the same way ENABLE_TOOL_SEARCH and the alias override already are. The autoCompactWindow key is still sent. Nothing changes when the setting is empty or the settings are a file path. Claude's real compaction point now matches the meter. No new regex or parser. Not run: a real Claude turn past 400k tokens (no Claude login on the VM, and very costly). That Claude Code obeys the variable over autoCompactWindow rests on the reporter's run of the real CLI 2.1.323.

Fixes #35725

Reproduction on main

On origin/main 48a56fa. (1) Real Claude Code 2.1.323 CLI in a scratch project whose .claude/settings.json sets env.CLAUDE_CODE_AUTO_COMPACT_WINDOW=1000000: claude -p /autocompact --settings '{"autoCompactWindow":400000}' (the shape T3 sends today) reported 'Auto-compact window for Opus 5.5: 1m tokens (from CLAUDE_CODE_AUTO_COMPACT_WINDOW)', so the T3 value is ignored. Adding "env":{"CLAUDE_CODE_AUTO_COMPACT_WINDOW":"400000"} to the same --settings reported '400k tokens'. (2) A unit test calling ClaudeAdapterV2.makeClaudeQueryOptions with settings.autoCompactWindow='400000' shows the launch settings carry no env.CLAUDE_CODE_AUTO_COMPACT_WINDOW (expected undefined to equal '400000'). I did not run a T3 thread past 400k tokens: there is no first-party Claude login on the VM, and such a turn is very costly.

Evidence

Set project env.CLAUDE_CODE_AUTO_COMPACT_WINDOW=1000000 and T3 Auto-compact after=400000. Generate launch settings through each commit's unmodified adapter, then run real Claude Code 2.1.323 with -p /autocompact --model claude-opus-5-5 --settings <generated JSON> on identical fresh synthetic projects on Linux.

Before	After
48a56faaa2	1fe034fc8a
CLI transcript: 1m tokens	CLI transcript: 400k tokens

Reproduction driver

This demonstrates real Claude settings precedence. Actual compaction and context-meter alignment remain unverified.

Not shown:

• Permanent blocker: actual compaction during a T3 Claude thread past 400k tokens requires first-party Claude authentication, which is unavailable on this VM. No actual compaction event or alignment with the approximately 367k context-meter tick was observed.
• The Settings UI and complete T3 server-to-provider launch were not exercised; the comparison directly called each commit's production adapter to generate settings and passed them to the real Claude CLI. Replay scenarios cannot demonstrate provider-specific settings precedence.

Tests

Confirmed the repro test fails on the base (expected undefined to equal '400000'). After the fix: vp test run src/orchestration-v2/Adapters/ClaudeAdapterV2.test.ts (apps/server): 406/406 passed, including the repro test. vp run check:changed: format, lint, knip, graphql, typecheck (3 packages) and tests (2 files, 408 tests) all passed.

Not in this change

• When the user's Claude settings are a file path (string), the existing code already replaces them with an object once autoCompactWindow is set; that behaviour is unchanged.

Who did the work

• Written by Claude Opus 5.5, in Claude Code
• Reviewed by GPT-6.1 Sol, in Codex, and by Claude Opus 5.5, each on its own
• Proof captured by GPT-6.1 Sol and checked by Claude Opus 5.5

Run by the t3code factory in T3 Code.

[macroscopeapp]

Approvability

Verdict: Approved at 1fe034f

Macroscope's review found this PR approvable — The adapter now sends an explicitly configured auto-compact window through Claude's higher-priority environment setting, so conflicting Claude configuration no longer overrides it. Empty settings and product defaults remain unchanged.

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

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ The exact PR base did not have a successful artifact. Baseline uses the latest successful main measurement shown below.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	21.5 KiB	21.5 KiB	0 B (0.0%)	24.4 KiB	✅
Codex	Live turn messages	1	1	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	0 B (0.0%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	0 B (0.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.9 KiB	21.9 KiB	0 B (0.0%)	24.4 KiB	✅
Claude	Live turn messages	1	1	0 (0.0%)	8	✅

Other scenarios

All 64 scenario measurements match main.

All 64 scenario measurements

Provider	Scenario	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	short turn resume	WebSocket decoded	10.5 KiB	10.5 KiB	0 B (0.0%)	12.7 KiB	✅
Codex	short turn resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	long turn running, cold open	HTTP decoded	281.7 KiB	281.7 KiB	0 B (0.0%)	302.7 KiB	✅
Codex	long turn running, cold open	rows	75	75	0 (0.0%)	75	✅
Codex	long turn running, history page	HTTP decoded	92.4 KiB	92.4 KiB	0 B (0.0%)	107.4 KiB	✅
Codex	long turn running, history page	rows	75	75	0 (0.0%)	75	✅
Codex	long turn finished, cold open	HTTP decoded	381.5 KiB	381.5 KiB	0 B (0.0%)	410.2 KiB	✅
Codex	long turn finished, cold open	rows	151	151	0 (0.0%)	151	✅
Codex	long turn finished, socket cold open	WebSocket decoded	381.6 KiB	381.6 KiB	0 B (0.0%)	410.2 KiB	✅
Codex	long turn finished, socket cold open	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	long turn finished, socket cold open	rows	151	151	0 (0.0%)	151	✅
Codex	long turn finished, work page	HTTP decoded	186.2 KiB	186.2 KiB	0 B (0.0%)	200.2 KiB	✅
Codex	long turn finished, work page	rows	150	150	0 (0.0%)	150	✅
Codex	idle thread resume	WebSocket decoded	65 B	65 B	0 B (0.0%)	200 B	✅
Codex	idle thread resume	WebSocket messages	1	1	0 (0.0%)	1	✅
Codex	fleet shell snapshot	HTTP decoded	42.4 KiB	42.4 KiB	0 B (0.0%)	53.7 KiB	✅
Codex	fleet shell snapshot	rows	27	27	0 (0.0%)	27	✅
Codex	fleet shell subscribe	WebSocket decoded	65 B	65 B	0 B (0.0%)	200 B	✅
Codex	fleet shell subscribe	WebSocket messages	1	1	0 (0.0%)	1	✅
Codex	fleet shell resume check	HTTP decoded	18 B	18 B	0 B (0.0%)	100 B	✅
Codex	fleet shell resume check	rows	0	0	0	0	✅
Codex	stale shell reload	HTTP decoded	42.4 KiB	42.4 KiB	0 B (0.0%)	53.7 KiB	✅
Codex	stale shell reload	rows	27	27	0 (0.0%)	27	✅
Codex	stale shell resume	WebSocket decoded	140 B	140 B	0 B (0.0%)	400 B	✅
Codex	stale shell resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Codex	stale todo list cold open	HTTP decoded	72.4 KiB	72.4 KiB	0 B (0.0%)	83.0 KiB	✅
Codex	stale todo list cold open	rows	75	75	0 (0.0%)	75	✅
Codex	stale todo list cold open	nodes	2	2	0 (0.0%)	4	✅
Codex	stale todo list cold open	subagents	2	2	0 (0.0%)	4	✅
Codex	stale todo list cold open	plans	1	1	0 (0.0%)	2	✅
Codex	stale todo list cold open	providerTurns	1	1	0 (0.0%)	2	✅
Codex	stale todo list cold open	checkpoints	1	1	0 (0.0%)	2	✅
Claude	short turn resume	WebSocket decoded	10.6 KiB	10.6 KiB	0 B (0.0%)	12.7 KiB	✅
Claude	short turn resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	long turn running, cold open	HTTP decoded	283.3 KiB	283.3 KiB	0 B (0.0%)	302.7 KiB	✅
Claude	long turn running, cold open	rows	75	75	0 (0.0%)	75	✅
Claude	long turn running, history page	HTTP decoded	93.4 KiB	93.4 KiB	0 B (0.0%)	107.4 KiB	✅
Claude	long turn running, history page	rows	75	75	0 (0.0%)	75	✅
Claude	long turn finished, cold open	HTTP decoded	384.2 KiB	384.2 KiB	0 B (0.0%)	410.2 KiB	✅
Claude	long turn finished, cold open	rows	151	151	0 (0.0%)	151	✅
Claude	long turn finished, socket cold open	WebSocket decoded	384.3 KiB	384.3 KiB	0 B (0.0%)	410.2 KiB	✅
Claude	long turn finished, socket cold open	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	long turn finished, socket cold open	rows	151	151	0 (0.0%)	151	✅
Claude	long turn finished, work page	HTTP decoded	188.1 KiB	188.1 KiB	0 B (0.0%)	200.2 KiB	✅
Claude	long turn finished, work page	rows	150	150	0 (0.0%)	150	✅
Claude	idle thread resume	WebSocket decoded	66 B	66 B	0 B (0.0%)	200 B	✅
Claude	idle thread resume	WebSocket messages	1	1	0 (0.0%)	1	✅
Claude	fleet shell snapshot	HTTP decoded	43.0 KiB	43.0 KiB	0 B (0.0%)	53.7 KiB	✅
Claude	fleet shell snapshot	rows	27	27	0 (0.0%)	27	✅
Claude	fleet shell subscribe	WebSocket decoded	66 B	66 B	0 B (0.0%)	200 B	✅
Claude	fleet shell subscribe	WebSocket messages	1	1	0 (0.0%)	1	✅
Claude	fleet shell resume check	HTTP decoded	18 B	18 B	0 B (0.0%)	100 B	✅
Claude	fleet shell resume check	rows	0	0	0	0	✅
Claude	stale shell reload	HTTP decoded	43.0 KiB	43.0 KiB	0 B (0.0%)	53.7 KiB	✅
Claude	stale shell reload	rows	27	27	0 (0.0%)	27	✅
Claude	stale shell resume	WebSocket decoded	140 B	140 B	0 B (0.0%)	400 B	✅
Claude	stale shell resume	WebSocket messages	2	2	0 (0.0%)	2	✅
Claude	stale todo list cold open	HTTP decoded	73.4 KiB	73.4 KiB	0 B (0.0%)	83.0 KiB	✅
Claude	stale todo list cold open	rows	75	75	0 (0.0%)	75	✅
Claude	stale todo list cold open	nodes	2	2	0 (0.0%)	4	✅
Claude	stale todo list cold open	subagents	2	2	0 (0.0%)	4	✅
Claude	stale todo list cold open	plans	1	1	0 (0.0%)	2	✅
Claude	stale todo list cold open	providerTurns	1	1	0 (0.0%)	2	✅
Claude	stale todo list cold open	checkpoints	1	1	0 (0.0%)	2	✅

Baseline: ac1cb92 · PR result: 1fe034f · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[coderabbitai]

No actionable comments were generated in the recent review.

Walkthrough

The Claude adapter now applies the configured auto-compact window to both SDK settings and CLAUDE_CODE_AUTO_COMPACT_WINDOW. A test verifies that the environment override receives the configured value.

Changes: Claude auto-compact configuration

Layer / File(s)	Summary
Apply and verify the configured window apps/server/src/orchestration-v2/Adapters/ClaudeAdapterV2.ts, apps/server/src/orchestration-v2/Adapters/ClaudeAdapterV2.test.ts	The adapter uses the configured nonempty window for the SDK option and, when settings are an object, the CLAUDE_CODE_AUTO_COMPACT_WINDOW environment override. String settings paths remain unchanged. The test checks that a configured value is passed through the environment override.

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

Change: Bug fix · Severity of issue fixed: Low

Possibly related issues

• pingdotgg/t3code#33798 concerns the same environment variable and compaction threshold. This change addresses which threshold Claude uses, but does not change how the context meter reads the variable.

Possibly related PRs

• pingdotgg/t3code#35726: It makes the same adapter change for the same issue.
• pingdotgg/t3code#29220: It resolves the adapter’s auto-compact setting and uses it for the context meter threshold, which relates to the value this PR passes to Claude.

Suggested reviewers: t3dotgg, maria-rcks, yash-singh1

Merge Risk: Minimal · up to 1fe03

The Claude auto-compact setting now takes precedence over a conflicting environment value in Claude settings. The behavior is unchanged when the setting is empty. No merge-blocking risk was found.

Pre-merge checks: 4

Check name	Status	Explanation
Title check		The title names the main change: Claude's auto-compact setting now takes precedence over CLAUDE_CODE_AUTO_COMPACT_WINDOW. The changeset sends the T3 value as env.CLAUDE_CODE_AUTO_COMPACT_WINDOW in the…
Linked Issues check		Issue #35725 requires that a filled "Auto-compact after" value take precedence over CLAUDE_CODE_AUTO_COMPACT_WINDOW in ~/.claude/settings.json. The issue names the fix: send the value as env.CLAUDE_CO…
Out of Scope Changes check		The whole-PR diff against merge base 48a56fa changes two files. ClaudeAdapterV2.ts changes only the auto-compact window path inside makeClaudeQueryOptions. The new env entry sits next to the existi…
Description check		The description covers the problem, the change, the scope, and the verification in detail. It explains the Claude Code precedence issue, the fix that sends CLAUDE_CODE_AUTO_COMPACT_WINDOW in the launc…

Full details: Title check

Explanation

The title names the main change: Claude's auto-compact setting now takes precedence over CLAUDE_CODE_AUTO_COMPACT_WINDOW. The changeset sends the T3 value as env.CLAUDE_CODE_AUTO_COMPACT_WINDOW in the launch settings, alongside autoCompactWindow, which matches the title's intent. The title uses the conventional commit format with a scope and is specific enough for a teammate scanning history. It is 88 characters, which is slightly long, but this does not reduce its clarity.

Full details: Linked Issues check

Explanation

Issue #35725 requires that a filled "Auto-compact after" value take precedence over CLAUDE_CODE_AUTO_COMPACT_WINDOW in ~/.claude/settings.json. The issue names the fix: send the value as env.CLAUDE_CODE_AUTO_COMPACT_WINDOW in --settings. At head 1fe034f, makeClaudeQueryOptions in apps/server/src/orchestration-v2/Adapters/ClaudeAdapterV2.ts reads autoCompactWindow once. When it is set and non-empty, the code adds CLAUDE_CODE_AUTO_COMPACT_WINDOW to the settings env map. It also keeps the autoCompactWindow key. The env merge follows the same pattern as ENABLE_TOOL_SEARCH. When the setting is empty or undefined, no env key is added. String settings paths pass through unchanged, as the issue scope allows. The PR adds a regression test in ClaudeAdapterV2.test.ts for the env value. The end-to-end effect in the real Claude Code CLI rests on the reporter's check with version 2.1.323. The author did not run a real turn past 400k tokens. That gap is a validation limit, not a missing coding requirement.

Full details: Out of Scope Changes check

Explanation

The whole-PR diff against merge base 48a56fa changes two files. ClaudeAdapterV2.ts changes only the auto-compact window path inside makeClaudeQueryOptions. The new env entry sits next to the existing alias and tool-search env handling. ClaudeAdapterV2.test.ts adds one regression test for the same behavior. No unrelated code paths changed. The file-path settings case is unchanged, as the PR description states.

Full details: Description check

Explanation

The description covers the problem, the change, the scope, and the verification in detail. It explains the Claude Code precedence issue, the fix that sends CLAUDE_CODE_AUTO_COMPACT_WINDOW in the launch settings env map, and the unchanged empty-setting and file-path behavior. It links the issue (#35725) and gives a reproduction on main, a before/after comparison, and unit test results. It also states what was not verified (a real Claude turn past 400k tokens). The template's Scope section asks for bug-fix framing, and the author links the bug issue, so that requirement is met. The verification section is thorough and names the models and harness used. The template's required Problem, Change, Scope, and Verification headings are not formally labeled, but the content maps onto each one. The description is mostly complete, so the check passes.

Finishing Touches

Generate unit tests (beta)

• Commit to this branch
• Create a new PR

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

Recent review info

Run configuration

• Source mode: Auto
• Repository: pingdotgg/t3code
• Branch: factory/issue-35725
• File: .coderabbit.config.ts
• Commit: 1fe034f
• Layers: Repository TS config
• Review profile: CHILL
• Plan: Team
• Run ID: 18f29ae3-fa9f-4a30-a379-02a652392cc1

Commits

Reviewing files that changed from the base of the PR and between c2931a6 and 1fe034f.

Files selected for processing: 2

• apps/server/src/orchestration-v2/Adapters/ClaudeAdapterV2.test.ts
• apps/server/src/orchestration-v2/Adapters/ClaudeAdapterV2.ts

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 1 remain after this review.