diff --git a/conductor/tracks/fix_test_failures_20260624/metadata.json b/conductor/tracks/fix_test_failures_20260624/metadata.json
new file mode 100644
index 00000000..ffba932f
--- /dev/null
+++ b/conductor/tracks/fix_test_failures_20260624/metadata.json
@@ -0,0 +1,59 @@
+{
+  "track_id": "fix_test_failures_20260624",
+  "name": "Fix 14 Test Failures (post-polish merge)",
+  "created_date": "2026-06-24",
+  "branch": "master",
+  "depends_on": ["code_path_audit_polish_20260622"],
+  "blocks": [],
+  "scope": {
+    "new_files": [],
+    "modified_files": [
+      "src/openai_schemas.py",
+      "tests/test_auto_whitelist.py",
+      "tests/test_command_palette_sim.py"
+    ],
+    "deleted_files": []
+  },
+  "estimated_effort": {
+    "method": "scope (per workflow.md §Tier 1 Track Initialization Rules). NO day estimates.",
+    "phase_1": "1 task: add custom __init__ to NormalizedResponse (fixes 12 tests)",
+    "phase_2": "1 task: update test_auto_whitelist_keywords to use dataclasses.replace",
+    "phase_3": "1 task: update test_palette_starts_hidden to use deterministic close",
+    "phase_4": "1 task: 6 verification criteria + TRACK_COMPLETION + state + tracks.md"
+  },
+  "verification_criteria": [
+    "VC1: 12 NormalizedResponse tests pass",
+    "VC2: test_auto_whitelist_keywords passes",
+    "VC3: test_palette_starts_hidden passes",
+    "VC4: full batched test suite is green (all 11 tiers PASS)",
+    "VC5: 4 audit gates remain clean",
+    "VC6: no new test failures introduced"
+  ],
+  "known_issues": [],
+  "deferred_to_followup_tracks": [],
+  "regressions_and_pre_existing_failures": [],
+  "pre_existing_failures_remaining": [],
+  "risk_register": [
+    {
+      "id": "risk-1",
+      "description": "Custom __init__ with init=False breaks other NormalizedResponse callers passing positional args",
+      "likelihood": "low",
+      "impact": "Phase 1 Task 1.1 may break production code at src/ai_client.py:908 or the 12 tests",
+      "mitigation": "All 12 tests + 1 production site use kwargs; verify with ast.parse + 12-test run"
+    },
+    {
+      "id": "risk-2",
+      "description": "dataclasses.replace for Session may not work if Session is slotted or has __slots__",
+      "likelihood": "low",
+      "impact": "Phase 2 Task 2.1 fails",
+      "mitigation": "Session is a regular @dataclass(frozen=True); replace() works on regular dataclasses"
+    },
+    {
+      "id": "risk-3",
+      "description": "A deterministic close callback does not exist in _predefined_callbacks",
+      "likelihood": "medium",
+      "impact": "Phase 3 Task 3.1 needs fallback (push toggle twice)",
+      "mitigation": "Plan includes the toggle-twice fallback in the Task 3.1 WHAT description"
+    }
+  ]
+}
\ No newline at end of file
diff --git a/conductor/tracks/fix_test_failures_20260624/plan.md b/conductor/tracks/fix_test_failures_20260624/plan.md
new file mode 100644
index 00000000..d76e5cc5
--- /dev/null
+++ b/conductor/tracks/fix_test_failures_20260624/plan.md
@@ -0,0 +1,104 @@
+# Plan: fix_test_failures_20260624
+
+3 phases, 3 tasks, 3 atomic commits. TDD: write the failing test invocation first (the existing 14 tests are the red), fix the code (green), commit.
+
+## Phase 1: NormalizedResponse dual-signature (1 task)
+
+Focus: Add custom `__init__` to `NormalizedResponse` that accepts both new nested `usage: UsageStats` and legacy flat kwargs. This fixes 12 of the 14 failing tests in one place.
+
+- [ ] Task 1.1: Add custom `__init__` to `NormalizedResponse` in `src/openai_schemas.py`.
+  - WHERE: `src/openai_schemas.py:75-80` (the `@dataclass(frozen=True) class NormalizedResponse:` block)
+  - WHAT:
+    - Change `@dataclass(frozen=True)` to `@dataclass(frozen=True, init=False)` on line 75
+    - Add a custom `__init__` method (after the class fields) that:
+      - Signature: `(self, text: str, tool_calls: tuple[ToolCall, ...] = (), usage: Optional[UsageStats] = None, raw_response: Any = None, usage_input_tokens: Optional[int] = None, usage_output_tokens: Optional[int] = None, usage_cache_read_tokens: Optional[int] = None, usage_cache_creation_tokens: Optional[int] = None) -> None`
+      - If `usage is None` and any legacy flat kwarg is non-None: build `UsageStats(input_tokens=usage_input_tokens or 0, output_tokens=usage_output_tokens or 0, cache_read_tokens=usage_cache_read_tokens or 0, cache_creation_tokens=usage_cache_creation_tokens or 0)`
+      - If `usage is None` and all legacy flat kwargs are None: build `UsageStats(0, 0)`
+      - Otherwise: use the provided `usage`
+      - Use `object.__setattr__(self, "text", text)` for all 4 field assignments (required because `frozen=True` locks `__setattr__`)
+  - HOW: Use `manual-slop_py_update_definition` to replace the entire class block. The new content includes the `init=False` decorator and the custom `__init__` method.
+  - SAFETY:
+    - Verify with `ast.parse(open("src/openai_schemas.py").read())` (no syntax errors)
+    - Run `uv run python -c "from src.openai_schemas import NormalizedResponse, UsageStats; r = NormalizedResponse(text='hi', tool_calls=(), usage_input_tokens=10, usage_output_tokens=5, raw_response=None); print(r.usage.input_tokens, r.usage.output_tokens)"`
+    - Run the 12 affected tests
+  - COMMIT: `fix(schemas): add legacy-kwarg backward compat to NormalizedResponse.__init__`
+  - GIT NOTE: 12 tests fixed by single backward-compat __init__; supports both new nested usage: UsageStats and legacy flat usage_input_tokens=... kwargs
+  - VERIFY: `uv run pytest tests/test_ai_client_cli.py tests/test_ai_client_tool_loop.py tests/test_ai_client_tool_loop_builder.py tests/test_ai_client_tool_loop_send_func.py tests/test_gemini_cli_integration.py tests/test_gemini_cli_parity_regression.py tests/test_gemini_cli_edge_cases.py -v` shows 12/12 pass
+
+## Phase 2: Session frozen fix (1 task)
+
+Focus: Update the test to use `dataclasses.replace()` instead of mutating the frozen `Session`.
+
+- [ ] Task 2.1: Update `test_auto_whitelist_keywords` to use `dataclasses.replace`.
+  - WHERE: `tests/test_auto_whitelist.py:20`
+  - WHAT: Change `reg.data[session_id]["whitelisted"] = True` to `reg.data[session_id] = dataclasses.replace(reg.data[session_id], whitelisted=True)`
+  - HOW: Use `manual-slop_edit_file` with the exact old/new text. Add `import dataclasses` to the top of the file if not already imported.
+  - SAFETY: Verify with `uv run pytest tests/test_auto_whitelist.py -v`
+  - COMMIT: `test(auto-whitelist): use dataclasses.replace for frozen Session mutation`
+  - GIT NOTE: 1 test fixed; the old `["whitelisted"] = True` was invalid on a frozen dataclass
+  - VERIFY: `uv run pytest tests/test_auto_whitelist.py -v` shows 0 failures
+
+## Phase 3: Toggle race fix (1 task)
+
+Focus: Replace the non-deterministic toggle with a deterministic close path.
+
+- [ ] Task 3.1: Update `test_palette_starts_hidden` to use a deterministic close callback.
+  - WHERE: `tests/test_command_palette_sim.py:38` (the `client.push_event("custom_callback", {"callback": "_toggle_command_palette", "args": []})` call)
+  - WHAT:
+    - Check `_predefined_callbacks` in `src/api_hooks.py` for a close callback (e.g., `_close_command_palette` or similar). If found, use it.
+    - If no close callback exists, push `_toggle_command_palette` once, poll for state, then push again to guarantee the palette is closed.
+  - HOW: Use `manual-slop_edit_file` with the exact old/new text.
+  - SAFETY: The test must end with `show_command_palette is False` regardless of starting state. Verify with `uv run pytest tests/test_command_palette_sim.py -v` in isolation AND in the full batch (per the `Isolated-Pass Verification Fallacy` rule in `conductor/workflow.md`).
+  - COMMIT: `test(palette): use deterministic close instead of toggle for palette_starts_hidden test`
+  - GIT NOTE: 1 test fixed; toggle is non-deterministic (close-if-open, open-if-closed); new code forces close regardless of starting state
+  - VERIFY: `uv run pytest tests/test_command_palette_sim.py -v` AND `uv run python scripts/run_tests_batched.py` (tier-3-live_gui)
+
+## Phase 4: Verification (1 task)
+
+Focus: Confirm all 10 verification criteria pass.
+
+- [ ] Task 4.1: Run all 6 VCs; write the track's end-of-track report.
+  - WHERE: Full test suite + audit gates
+  - WHAT:
+    - Run VC1-VC6 (the 6 verification criteria from the spec)
+    - Write `docs/reports/TRACK_COMPLETION_fix_test_failures_20260624.md` with the verification results
+    - Update this track's `state.toml` to `status = "completed"`, `current_phase = "complete"`, all 4 phases `completed`
+    - Update `conductor/tracks.md` to add a row for this track
+  - HOW: Run each VC command, capture output, write the report.
+  - SAFETY: The 2 pre-existing-violation audit gates (NG1, NG2 from the polish track) are still out of scope. Do not regress them.
+  - COMMIT: 3 commits: `conductor(state): fix_test_failures_20260624 SHIPPED`, `docs(reports): TRACK_COMPLETION for fix_test_failures_20260624`, `conductor(tracks): add fix_test_failures_20260624 row`
+  - GIT NOTE: 1 per commit per workflow.md
+  - VERIFY: All 6 VCs pass; the 14 previously-failing tests are now green; no new failures introduced
+
+## Commit Log (Expected)
+
+1. `fix(schemas): add legacy-kwarg backward compat to NormalizedResponse.__init__` (Task 1.1)
+2. `test(auto-whitelist): use dataclasses.replace for frozen Session mutation` (Task 2.1)
+3. `test(palette): use deterministic close instead of toggle for palette_starts_hidden test` (Task 3.1)
+4. `conductor(state): fix_test_failures_20260624 SHIPPED` (Task 4.1)
+5. `docs(reports): TRACK_COMPLETION for fix_test_failures_20260624` (Task 4.1)
+6. `conductor(tracks): add fix_test_failures_20260624 row` (Task 4.1)
+
+## Verification Commands (run at end of Phase 4)
+
+```bash
+# VC1: the 12 NormalizedResponse tests pass
+uv run pytest tests/test_ai_client_cli.py tests/test_ai_client_tool_loop.py tests/test_ai_client_tool_loop_builder.py tests/test_ai_client_tool_loop_send_func.py tests/test_gemini_cli_integration.py tests/test_gemini_cli_parity_regression.py tests/test_gemini_cli_edge_cases.py -v
+
+# VC2: test_auto_whitelist_keywords passes
+uv run pytest tests/test_auto_whitelist.py -v
+
+# VC3: test_palette_starts_hidden passes
+uv run pytest tests/test_command_palette_sim.py -v
+
+# VC4: full batched suite is green
+uv run python scripts/run_tests_batched.py
+
+# VC5: 4 audit gates
+uv run python scripts/audit_exception_handling.py
+uv run python scripts/audit_weak_types.py
+uv run python scripts/audit_main_thread_imports.py
+uv run python scripts/audit_no_models_config_io.py
+
+# VC6: confirm no new failures (diff against pre-fix summary)
+```
\ No newline at end of file
diff --git a/conductor/tracks/fix_test_failures_20260624/spec.md b/conductor/tracks/fix_test_failures_20260624/spec.md
new file mode 100644
index 00000000..9eea35e1
--- /dev/null
+++ b/conductor/tracks/fix_test_failures_20260624/spec.md
@@ -0,0 +1,133 @@
+# Track Specification: fix_test_failures_20260624
+
+## Overview
+
+3 surgical fixes to make the full test suite green after the polish merge. The 14 failing tests cluster into 3 root causes; each is a single-file change.
+
+## Current State Audit (commit 4c2bb3c9, master)
+
+3 tier failures from `uv run python scripts/run_tests_batched.py`:
+
+### Root cause 1: NormalizedResponse API mismatch (12 tests)
+
+`src/openai_schemas.py:75-80` defines `NormalizedResponse` with 4 fields: `text`, `tool_calls`, `usage: UsageStats`, `raw_response`. The `@dataclass` auto-generates `__init__` requiring `usage: UsageStats`.
+
+But 12 tests + 1 production site at `src/ai_client.py:908` call it with the OLD flat-kwarg API:
+```python
+NormalizedResponse(
+    text=..., tool_calls=...,
+    usage_input_tokens=10, usage_output_tokens=5,
+    usage_cache_read_tokens=0, usage_cache_creation_tokens=0,
+    raw_response=None,
+)
+```
+
+**Failing tests:**
+- `tests/test_ai_client_cli.py::test_ai_client_send_gemini_cli`
+- `tests/test_ai_client_tool_loop.py` (5 tests)
+- `tests/test_ai_client_tool_loop_builder.py::test_run_with_tool_loop_calls_request_builder_each_round`
+- `tests/test_ai_client_tool_loop_send_func.py` (2 tests)
+- `tests/test_gemini_cli_integration.py::test_gemini_cli_full_integration`
+- `tests/test_gemini_cli_parity_regression.py::test_send_invokes_adapter_send`
+- `tests/test_gemini_cli_edge_cases.py::test_gemini_cli_loop_termination`
+
+**Fix:** add a custom `__init__` to `NormalizedResponse` (using `@dataclass(frozen=True, init=False)`) that accepts BOTH signatures. The legacy flat kwargs are auto-converted to a `UsageStats` instance.
+
+### Root cause 2: Session dataclass frozen (1 test)
+
+`tests/test_auto_whitelist.py:20` does `reg.data[session_id]["whitelisted"] = True`. `Session` is `@dataclass(frozen=True)` so attribute assignment fails.
+
+**Failing test:** `tests/test_auto_whitelist.py::test_auto_whitelist_keywords`
+
+**Fix:** update the test to use `dataclasses.replace(reg.data[session_id], whitelisted=True)` (or equivalent setter on the registry).
+
+### Root cause 3: Toggle race (1 test)
+
+`tests/test_command_palette_sim.py::test_palette_starts_hidden` calls `_toggle_command_palette` (non-deterministic) and asserts `show_command_palette is False`. The toggle is open-if-closed, close-if-open, so the assertion depends on prior state.
+
+**Failing test:** `tests/test_command_palette_sim.py::test_palette_starts_hidden`
+
+**Fix:** call a deterministic close callback (e.g., `_close_command_palette`) instead of toggle. If no such callback exists, push the toggle twice (open then close).
+
+## Goals
+
+| ID | Goal | Acceptance |
+|---|---|---|
+| G1 | `NormalizedResponse` accepts both the new `usage: UsageStats` and the legacy flat `usage_input_tokens=...` kwargs | All 12 affected tests pass; existing usage in `src/ai_client.py:908` continues to work |
+| G2 | `test_auto_whitelist_keywords` no longer attempts to mutate a frozen `Session` | The test passes; no production code change |
+| G3 | `test_palette_starts_hidden` uses a deterministic close path | The test passes; the test no longer depends on prior fixture state |
+| G4 | Full test suite (`scripts/run_tests_batched.py`) is green | 0 failed, 0 new failures vs baseline |
+
+## Non-Goals
+
+- Refactoring `NormalizedResponse` to remove the legacy support (the fix is additive; legacy kwargs continue to work)
+- Migrating `src/ai_client.py:908` from the legacy API to the new API (out of scope; the fix makes both work)
+- Adding new tests beyond fixing the 14 failing ones
+- Touching any other files
+
+## Functional Requirements
+
+### FR1: NormalizedResponse dual-signature __init__
+
+`src/openai_schemas.py` line 75-80: change `@dataclass(frozen=True)` to `@dataclass(frozen=True, init=False)` and add a custom `__init__` that:
+- Accepts all 4 canonical kwargs: `text`, `tool_calls`, `usage`, `raw_response` (with defaults)
+- Accepts all 4 legacy flat kwargs: `usage_input_tokens`, `usage_output_tokens`, `usage_cache_read_tokens`, `usage_cache_creation_tokens` (all `Optional[int]`, default `None`)
+- If `usage` is None AND any legacy flat kwarg is non-None, builds a `UsageStats` from the legacy kwargs
+- If both `usage` and legacy kwargs are None, builds `UsageStats(0, 0)`
+- Uses `object.__setattr__` for all 4 field assignments (required because `frozen=True`)
+
+### FR2: test_auto_whitelist_keywords uses replace
+
+`tests/test_auto_whitelist.py:20`: change `reg.data[session_id]["whitelisted"] = True` to use `dataclasses.replace(reg.data[session_id], whitelisted=True)` and assign back to `reg.data[session_id]`.
+
+### FR3: test_palette_starts_hidden uses deterministic close
+
+`tests/test_command_palette_sim.py:38`: change the `_toggle_command_palette` callback to a deterministic close. Options (in preference order):
+- (a) Call `_close_command_palette` directly if it exists in `_predefined_callbacks`
+- (b) If no close callback exists, push `_toggle_command_palette` once, wait for state to settle, then push it again to guarantee close
+- The test must end with `show_command_palette is False` regardless of starting state
+
+## Non-Functional Requirements
+
+- NFR1: 1-space indentation
+- NFR2: CRLF line endings (Windows)
+- NFR3: No comments in source code
+- NFR4: Per-task atomic commits
+- NFR5: No new pip dependencies
+- NFR6: 0 new test failures introduced
+
+## Architecture Reference
+
+- `conductor/code_styleguides/error_handling.md` — `Result[T]` convention (no changes needed; no Result changes in this fix)
+- `conductor/code_styleguides/data_oriented_design.md` — DOD reference (the NormalizedResponse fix is purely additive)
+- `src/openai_schemas.py:75-93` — the file being modified (FR1)
+- `tests/test_auto_whitelist.py:20` — the file being modified (FR2)
+- `tests/test_command_palette_sim.py:38` — the file being modified (FR3)
+
+## Out of Scope
+
+- Migrating `src/ai_client.py:908` from the legacy API to `usage: UsageStats(...)` (the additive fix makes both work; explicit migration is a separate concern)
+- Adding regression tests for the new NormalizedResponse dual-signature behavior (the 12 existing tests ARE the regression test)
+- Refactoring Session to not be frozen (the test fix is simpler)
+- Fixing the root cause of `_toggle_command_palette` being non-deterministic (the test fix is simpler)
+- Any changes to test infrastructure or conftest
+
+## Verification Criteria (Definition of Done)
+
+| # | Criterion | Verification command |
+|---|---|---|
+| VC1 | The 12 NormalizedResponse tests pass | `uv run pytest tests/test_ai_client_cli.py tests/test_ai_client_tool_loop.py tests/test_ai_client_tool_loop_builder.py tests/test_ai_client_tool_loop_send_func.py tests/test_gemini_cli_integration.py tests/test_gemini_cli_parity_regression.py tests/test_gemini_cli_edge_cases.py -v` |
+| VC2 | `test_auto_whitelist_keywords` passes | `uv run pytest tests/test_auto_whitelist.py -v` |
+| VC3 | `test_palette_starts_hidden` passes | `uv run pytest tests/test_command_palette_sim.py -v` |
+| VC4 | Full batched test suite is green | `uv run python scripts/run_tests_batched.py` (all 11 tiers PASS) |
+| VC5 | The 4 mandatory audit gates remain clean | `uv run python scripts/audit_exception_handling.py` (informational); `uv run python scripts/audit_weak_types.py`; `uv run python scripts/audit_main_thread_imports.py`; `uv run python scripts/audit_no_models_config_io.py` |
+| VC6 | No new test failures introduced | diff `uv run python scripts/run_tests_batched.py` summary table against pre-fix baseline (the 14 known failures are now green; no new failures added) |
+
+## Risks
+
+| # | Risk | Likelihood | Mitigation |
+|---|---|---|---|
+| R1 | The custom `__init__` with `init=False` breaks other NormalizedResponse callers that pass positional args | low | The class is only constructed in 1 production site (`src/ai_client.py:908`) and the 12 test sites, all of which use kwargs. Verify by running the 12 tests. |
+| R2 | The `dataclasses.replace()` for Session assignment to `reg.data[session_id]` mutates the registry's internal state in a way that breaks other tests | low | The registry stores sessions in a dict. `reg.data[session_id] = replace(...)` is a normal dict assignment, equivalent to the old `reg.data[session_id]["whitelisted"] = True` semantics. |
+| R3 | A deterministic close callback doesn't exist in `_predefined_callbacks` | medium | Fallback option: push toggle twice. Verify by checking the available callbacks in the API hooks registry. |
+| R4 | `UsageStats(0, 0)` syntax (when no kwargs and no legacy) is wrong if UsageStats requires explicit args | low | Verified: `UsageStats` has `input_tokens: int` and `output_tokens: int` as required, with `cache_*_tokens: int = 0` defaults. `UsageStats(0, 0)` is valid. |
\ No newline at end of file
diff --git a/conductor/tracks/fix_test_failures_20260624/state.toml b/conductor/tracks/fix_test_failures_20260624/state.toml
new file mode 100644
index 00000000..15ca7ee5
--- /dev/null
+++ b/conductor/tracks/fix_test_failures_20260624/state.toml
@@ -0,0 +1,39 @@
+# Track state for fix_test_failures_20260624
+# 3 surgical fixes for 14 post-polish-merge test failures.
+# 4 phases, 4 tasks. Tier 2 to execute per conductor/workflow.md.
+
+[meta]
+track_id = "fix_test_failures_20260624"
+name = "Fix 14 Test Failures (post-polish merge)"
+status = "active"
+current_phase = 0
+last_updated = "2026-06-24"
+
+[parent]
+# Follow-up to code_path_audit_polish_20260622 (merged)
+
+[blocked_by]
+code_path_audit_polish_20260622 = "merged"
+
+[blocks]
+# This track blocks nothing. It is a test-fix task.
+
+[phases]
+phase_1 = { status = "pending", checkpointsha = "", name = "NormalizedResponse dual-signature __init__" }
+phase_2 = { status = "pending", checkpointsha = "", name = "Session frozen fix in test_auto_whitelist" }
+phase_3 = { status = "pending", checkpointsha = "", name = "Toggle race fix in test_palette_starts_hidden" }
+phase_4 = { status = "pending", checkpointsha = "", name = "Verification + End-of-Track Report" }
+
+[tasks]
+t1_1 = { status = "pending", commit_sha = "", description = "Add custom __init__ to NormalizedResponse accepting both nested usage: UsageStats and legacy flat usage_input_tokens=... kwargs" }
+t2_1 = { status = "pending", commit_sha = "", description = "Update test_auto_whitelist_keywords to use dataclasses.replace for frozen Session mutation" }
+t3_1 = { status = "pending", commit_sha = "", description = "Update test_palette_starts_hidden to use deterministic close callback instead of non-deterministic toggle" }
+t4_1 = { status = "pending", commit_sha = "", description = "Run all 6 VCs; write TRACK_COMPLETION report; update this state.toml + conductor/tracks.md" }
+
+[verification]
+vc1_normalized_response_tests_pass = false
+vc2_auto_whitelist_test_passes = false
+vc3_palette_starts_hidden_test_passes = false
+vc4_full_batched_suite_green = false
+vc5_audit_gates_clean = false
+vc6_no_new_test_failures = false
\ No newline at end of file