fix(ci): repair the Windows probe test and actually run repository suites

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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

[rynfar]

scripts/build-desktop-artifact.test.ts has been failing on pylon:

× skips the primary native probe for cross-architecture Windows payloads
AssertionError: expected true to be false

The test was wrong, not the code

It asserted that no spawned command carried ELECTRON_RUN_AS_NODE=1. But
verifyWindowsPrimaryFffNativeLoad spreads process.env into the child, so that
assertion trips whenever the runner itself has the variable set — which is the
case under any Electron-hosted shell. The arch gate was skipping the probe
correctly the whole time; the test was reading an inherited environment variable
and calling it evidence.

It now asserts what it means: the packaged executable was never spawned.
That is the only thing the probe does, and it cannot be faked by inheritance.

Why nobody noticed

CI never ran the file. The Test job filters out @t3tools/monorepo, whose own
test script is vp run -r test and would recurse — so the root package's
suites, scripts/ and the oxlint plugin, 396 tests, ran nowhere at all.

Adds a step that runs them directly, plus a vitest exclusion for a plain node
.cjs helper under .github/scripts/ that vitest would otherwise collect and
fail on with "No test suite found in file".

Verification

• vp test run scripts/build-desktop-artifact.test.ts — 68 passed (was 67/1)
• vp test run scripts oxlint-plugin-t3code — 396 passed, 35 files, so the
  new CI step is green today
• Mutation probe: removing the arch gate
  (hostArchitecture !== input.targetArch) still fails the cross-architecture
  test, so the retargeted assertion is load-bearing rather than merely passing

Closes#147.

Claude Opus 5 via Claude Code.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	+18 B (+0.1%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+2 B (+0.0%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+16 B (+0.2%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	0 B (0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	−1 B (−0.0%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	+1 B (+0.0%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.4 KiB	56.4 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	10	10	0 (0.0%)	21	✅

Baseline: 71a06e0 · PR result: 5678b0e · 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: 109.4 KiB
• Claude decoded thread snapshot: 110.1 KiB

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