SoundBlaster · SoundBlaster · Feb 18, 2026 · Feb 18, 2026 · Feb 18, 2026 · Feb 18, 2026
diff --git a/SPECS/ARCHIVE/INDEX.md b/SPECS/ARCHIVE/INDEX.md
@@ -1,6 +1,6 @@
 # mcpbridge-wrapper Tasks Archive
 
-**Last Updated:** 2026-02-18 (P13-T4)
+**Last Updated:** 2026-02-18 (P13-T5)
 
 ## Archived Tasks
 
@@ -108,6 +108,7 @@
 | P13-T2 | [P13-T2_Implement_persistent_broker_daemon/](P13-T2_Implement_persistent_broker_daemon/) | 2026-02-17 | PASS |
 | P13-T3 | [P13-T3_Implement_multi-client_transport_and_JSON-RPC_multiplexing/](P13-T3_Implement_multi-client_transport_and_JSON-RPC_multiplexing/) | 2026-02-18 | PASS |
 | P13-T4 | [P13-T4_Add_stdio_proxy_mode/](P13-T4_Add_stdio_proxy_mode/) | 2026-02-18 | PASS |
+| P13-T5 | [P13-T5_Validate_prompt_reduction_and_multi_client_stability/](P13-T5_Validate_prompt_reduction_and_multi_client_stability/) | 2026-02-18 | PARTIAL |
 
 ## Historical Artifacts
 
@@ -179,6 +180,7 @@
 | [REVIEW_P13-T2_broker_daemon.md](_Historical/REVIEW_P13-T2_broker_daemon.md) | Review report for P13-T2 |
 | [REVIEW_P13-T3_transport_multiplexing.md](_Historical/REVIEW_P13-T3_transport_multiplexing.md) | Review report for P13-T3 |
 | [REVIEW_P13-T4_stdio_proxy_mode.md](_Historical/REVIEW_P13-T4_stdio_proxy_mode.md) | Review report for P13-T4 |
+| [REVIEW_P13-T5_prompt_reduction_multi_client_stability.md](_Historical/REVIEW_P13-T5_prompt_reduction_multi_client_stability.md) | Review report for P13-T5 |
 
 ## Archive Log
 
@@ -312,3 +314,5 @@
 | 2026-02-18 | P13-T3 | Archived REVIEW_P13-T3_transport_multiplexing report |
 | 2026-02-18 | P13-T4 | Archived Add_stdio_proxy_mode (PASS) |
 | 2026-02-18 | P13-T4 | Archived REVIEW_P13-T4_stdio_proxy_mode report |
+| 2026-02-18 | P13-T5 | Archived Validate_prompt_reduction_and_multi_client_stability (PARTIAL) |
+| 2026-02-18 | P13-T5 | Archived REVIEW_P13-T5_prompt_reduction_multi_client_stability report |
diff --git a/...client_stability/P13-T5_Validate_prompt_reduction_and_multi_client_stability.md b/...client_stability/P13-T5_Validate_prompt_reduction_and_multi_client_stability.md
@@ -0,0 +1,95 @@
+# PRD: P13-T5 — Validate prompt reduction and multi-client stability
+
+**Status:** IN PROGRESS
+**Priority:** P1
+**Branch:** `feature/P13-T5-prompt-reduction-multi-client-stability`
+**Depends on:** P13-T4 ✅
+
+---
+
+## 1. Overview
+
+P13-T4 introduced broker proxy mode so short-lived MCP client processes can forward through one long-lived broker session. P13-T5 validates that behavior under realistic usage patterns and captures evidence that broker mode reduces upstream bridge churn (and therefore reduces repeated Xcode permission prompts).
+
+This task delivers a reproducible integration test suite, a process-churn comparison report, and a manual validation report that confirms observed prompt behavior while the broker remains running.
+
+---
+
+## 2. Scope
+
+### In-scope
+- Add integration tests for sequential and concurrent short-lived proxy clients that share one broker-owned upstream bridge.
+- Add assertions/evidence capture for upstream bridge lifecycle stability during client churn.
+- Add manual validation report documenting Xcode permission prompt behavior in direct mode vs broker mode.
+- Add regression validation commands and results to the task validation report.
+
+### Out-of-scope
+- Broker mode documentation rollout/migration guides (P13-T6).
+- New broker runtime features unrelated to validation.
+- Non-Unix transport behavior.
+
+---
+
+## 3. Design
+
+### 3.1 Test strategy
+
+Create `tests/integration/test_broker_multi_client.py` with targeted scenarios:
+
+1. Sequential short-lived clients:
+   - Launch N short-lived proxy client sessions against one broker.
+   - Verify all sessions complete successfully.
+   - Verify upstream bridge process identity/count stays stable.
+
+2. Concurrent client load:
+   - Launch M concurrent client sessions.
+   - Verify responses are routed correctly and no cross-talk/corruption occurs.
+   - Verify no unexpected broker/upstream restarts during load.
+
+### 3.2 Metrics artifact
+
+Create a metrics artifact in the task archive that compares:
+- Direct mode: upstream process starts for N short-lived sessions.
+- Broker mode: upstream process starts for N short-lived sessions.
+
+The artifact should include commands/inputs, observed counts, and interpretation.
+
+### 3.3 Manual validation artifact
+
+Create a manual validation report that records:
+- Environment (macOS, Python, Xcode version)
+- Steps to reproduce for direct mode and broker mode
+- Prompt observations for both modes
+- Result verdict aligned to acceptance criteria
+
+---
+
+## 4. File changes
+
+| File | Change |
+|------|--------|
+| `tests/integration/test_broker_multi_client.py` | Add broker multi-client stability integration tests |
+| `SPECS/INPROGRESS/P13-T5_Validation_Report.md` | Record quality gate outcomes and acceptance-criteria evidence |
+| `SPECS/INPROGRESS/P13-T5_manual_prompt_validation.md` | Manual direct-vs-broker prompt behavior notes |
+
+---
+
+## 5. Acceptance criteria
+
+- [ ] Sequential short-lived clients reuse one broker-owned upstream bridge process
+- [ ] Concurrent client tool calls remain stable under load
+- [ ] Manual test confirms no extra Xcode prompt while broker stays running
+- [ ] Regression suite passes with broker mode enabled
+
+---
+
+## 6. Quality gates
+
+- `pytest` — all tests pass
+- `ruff check src/ tests/` — no lint errors
+- `mypy src/` — no new type errors
+- `pytest --cov` — coverage ≥ 90%
+
+---
+**Archived:** 2026-02-18
+**Verdict:** PARTIAL
diff --git a/...alidate_prompt_reduction_and_multi_client_stability/P13-T5_Validation_Report.md b/...alidate_prompt_reduction_and_multi_client_stability/P13-T5_Validation_Report.md
@@ -0,0 +1,44 @@
+# Validation Report: P13-T5 — Validate prompt reduction and multi-client stability
+
+**Date:** 2026-02-18
+**Branch:** `feature/P13-T5-prompt-reduction-multi-client-stability`
+**Verdict:** PARTIAL
+
+---
+
+## Quality Gates
+
+| Gate | Result | Details |
+|------|--------|---------|
+| `pytest` | ✅ PASS | 577 passed, 5 skipped |
+| `ruff check src/ tests/` | ✅ PASS | All checks passed |
+| `mypy src/` | ✅ PASS | Success: no issues found in 18 source files |
+| `pytest --cov` ≥ 90% | ✅ PASS | 92.31% total coverage |
+
+---
+
+## Acceptance Criteria
+
+| Criterion | Status | Evidence |
+|-----------|--------|----------|
+| Sequential short-lived clients reuse one broker-owned upstream bridge process | ✅ | `tests/integration/test_broker_multi_client.py::test_sequential_short_lived_clients_reuse_single_upstream_bridge` |
+| Concurrent client tool calls remain stable under load | ✅ | `tests/integration/test_broker_multi_client.py::test_concurrent_clients_remain_stable_under_load` |
+| Manual test confirms no extra Xcode prompt while broker stays running | ⚠️ PARTIAL | `SPECS/INPROGRESS/P13-T5_manual_prompt_validation.md` (interactive prompt observation pending human-run desktop check) |
+| Regression suite passes with broker mode enabled | ✅ | Full `pytest` and `pytest --cov` runs pass |
+
+---
+
+## Artifacts
+
+| File | Description |
+|------|-------------|
+| `tests/integration/test_broker_multi_client.py` | New integration coverage for sequential reuse, concurrent stability, and single-upstream launch count |
+| `SPECS/INPROGRESS/P13-T5_process_churn_metrics.md` | Direct-vs-broker upstream process churn comparison |
+| `SPECS/INPROGRESS/P13-T5_manual_prompt_validation.md` | Manual prompt validation procedure and current status |
+
+---
+
+## Notes
+
+- Process churn evidence shows broker mode reduced upstream process starts from 12 to 1 for equivalent short-lived session count in local validation.
+- The remaining gap to a full PASS verdict is interactive prompt observation in an operator-driven desktop run.
diff --git a/..._prompt_reduction_and_multi_client_stability/P13-T5_manual_prompt_validation.md b/..._prompt_reduction_and_multi_client_stability/P13-T5_manual_prompt_validation.md
@@ -0,0 +1,30 @@
+# Manual Prompt Validation: P13-T5
+
+**Date:** 2026-02-18
+**Task:** Validate reduced Xcode permission prompt behavior in broker mode
+
+## Environment checks
+
+- `xcrun mcpbridge --help` executed successfully.
+- Xcode process detected (`pgrep -x Xcode` returned a running PID during validation).
+
+## Manual procedure
+
+1. Start from a clean state (no stale broker processes/sockets).
+2. Run repeated short-lived sessions in direct mode and record prompt behavior.
+3. Run repeated short-lived sessions in broker mode and record prompt behavior.
+4. Confirm whether prompts reappear while the broker-owned upstream session remains running.
+
+## Result
+
+**Status:** ⚠️ PARTIAL
+
+- Automated evidence confirms broker mode keeps a single upstream process across many short-lived sessions.
+- Interactive macOS prompt observation could not be fully captured from this non-interactive terminal workflow.
+- A human-operated verification pass in an interactive desktop session is still required to conclusively mark prompt behavior as PASS.
+
+## Supporting automated evidence
+
+- `tests/integration/test_broker_multi_client.py` covers sequential reuse and concurrent stability.
+- `test_broker_mode_launches_upstream_once_for_many_short_lived_clients` verifies a single upstream launch across 12 short-lived sessions.
+- `SPECS/INPROGRESS/P13-T5_process_churn_metrics.md` captures direct-vs-broker churn comparison.
diff --git a/...ate_prompt_reduction_and_multi_client_stability/P13-T5_process_churn_metrics.md b/...ate_prompt_reduction_and_multi_client_stability/P13-T5_process_churn_metrics.md
@@ -0,0 +1,62 @@
+# P13-T5 Process Churn Metrics (Direct vs Broker)
+
+**Date:** 2026-02-18
+
+## Summary
+
+Measured process churn for 12 short-lived sessions:
+
+| Mode | Sessions | Upstream process starts | Notes |
+|------|----------|-------------------------|-------|
+| Direct (per-session subprocess baseline) | 12 | 12 | Measured via local Python harness that starts one upstream stub per client session |
+| Broker mode (P13 transport/proxy architecture) | 12 | 1 | Validated by `test_broker_mode_launches_upstream_once_for_many_short_lived_clients` |
+
+## Evidence
+
+### Direct-mode baseline command
+
+```bash
+python - <<'PY'
+import subprocess
+import sys
+import tempfile
+from pathlib import Path
+
+with tempfile.TemporaryDirectory(prefix='p13t5-') as tmp:
+    script = Path(tmp) / 'direct_mode_stub.py'
+    script.write_text(
+        'import sys\n'
+        'for _ in sys.stdin:\n'
+        '    pass\n'
+    )
+
+    pids = set()
+    for _ in range(12):
+        proc = subprocess.Popen(
+            [sys.executable, '-u', str(script)],
+            stdin=subprocess.PIPE,
+            stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE,
+            text=True,
+        )
+        proc.communicate(input='{"jsonrpc":"2.0","id":1}\n', timeout=2)
+        pids.add(proc.pid)
+
+    print(f'direct_mode_process_starts={len(pids)}')
+PY
+```
+
+Output:
+
+```text
+direct_mode_process_starts=12
+```
+
+### Broker-mode evidence
+
+`tests/integration/test_broker_multi_client.py::test_broker_mode_launches_upstream_once_for_many_short_lived_clients`
+asserts `launch_count == 1` after 12 short-lived client sessions.
+
+## Interpretation
+
+For equivalent short-lived session count (N=12), broker mode reduced upstream starts from 12 to 1 (91.7% reduction), which directly addresses upstream churn that contributes to repeated authorization prompts.
diff --git a/SPECS/ARCHIVE/_Historical/REVIEW_P13-T5_prompt_reduction_multi_client_stability.md b/SPECS/ARCHIVE/_Historical/REVIEW_P13-T5_prompt_reduction_multi_client_stability.md
@@ -0,0 +1,30 @@
+## REVIEW REPORT — P13-T5 prompt reduction and multi-client stability
+
+**Scope:** origin/main..HEAD
+**Files:** 8
+
+### Summary Verdict
+- [ ] Approve
+- [x] Approve with comments
+- [ ] Request changes
+- [ ] Block
+
+### Critical Issues
+- None.
+
+### Secondary Issues
+- [Low] Interactive Xcode permission prompt verification remains manual-only and is recorded as PARTIAL in the validation artifact; this is a known limitation of non-interactive automation, not a code defect.
+
+### Architectural Notes
+- The new integration tests exercise the real broker daemon + Unix socket transport + subprocess upstream path and verify both sequential reuse and concurrent routing stability.
+- Process churn metrics are explicit and reproducible, with direct-vs-broker comparison captured in archive artifacts.
+
+### Tests
+- `pytest` — 577 passed, 5 skipped
+- `ruff check src/ tests/` — pass
+- `mypy src/` — pass
+- `pytest --cov` — 92.31% total (>=90%)
+
+### Next Steps
+- No new follow-up tasks required from this review.
+- Existing P13-T6 remains the next documentation/migration task.
diff --git a/SPECS/INPROGRESS/next.md b/SPECS/INPROGRESS/next.md
@@ -1,9 +1,8 @@
 # No Active Task
 
-The previously selected task has been archived.
-
 ## Recently Archived
 
+- 2026-02-18 — P13-T5: Validate prompt reduction and multi-client stability (PARTIAL)
 - 2026-02-18 — P13-T4: Add stdio proxy mode for compatibility with existing MCP clients (PASS)
 - 2026-02-18 — P13-T3: Implement multi-client transport and JSON-RPC multiplexing (PASS)
 - 2026-02-17 — P13-T2: Implement persistent broker daemon with single upstream Xcode bridge (PASS)
@@ -12,6 +11,6 @@ The previously selected task has been archived.
 
 ## Suggested Next Tasks
 
-- P13-T5: Validate prompt reduction and multi-client stability (P1, depends on P13-T4 ✅)
-- P13-T6: Document broker mode configuration, migration, and rollback (P1, depends on P13-T4 ✅)
-- FU-P12-T1-1: Remove or document `MCPInitializeParams` in schemas (P3)
+- P13-T6 — Document broker mode configuration, migration, and rollback (P1, depends on P13-T4, partially unblocked by P13-T5 artifacts)
+- FU-P13-T4-1 — Replace deprecated `asyncio.get_event_loop()` calls in `broker/proxy.py` (P2)
+- FU-P13-T4-2 — Implement or remove `reconnect` parameter in `BrokerProxy` (P2)
diff --git a/SPECS/Workplan.md b/SPECS/Workplan.md
@@ -1094,7 +1094,7 @@ Keep a single long-lived client/session running to reduce process churn. This is
 #### Resolution Path
 - [x] Design persistent broker architecture for shared upstream Xcode session (P13-T1)
 - [x] Implement long-lived broker daemon with single upstream bridge connection (P13-T2)
-- [ ] Add multi-client transport + stdio proxy mode to reuse broker session (P13-T3, P13-T4)
+- [x] Add multi-client transport + stdio proxy mode to reuse broker session (P13-T3, P13-T4)
 - [ ] Validate reduced prompt behavior and document rollout/migration steps (P13-T5, P13-T6)
 
 ---
@@ -2094,7 +2094,8 @@ Phase 9 Follow-up Backlog
 
 ---
 
-#### P13-T5: Validate prompt reduction and multi-client stability
+#### ✅ P13-T5: Validate prompt reduction and multi-client stability
+- **Status:** ⚠️ PARTIAL (2026-02-18, interactive prompt verification pending)
 - **Description:** Add integration and manual verification that repeated short-lived client sessions can reuse the broker session without repeated upstream churn, plus load tests for concurrent calls.
 - **Priority:** P1
 - **Dependencies:** P13-T4
@@ -2104,10 +2105,10 @@ Phase 9 Follow-up Backlog
   - Manual validation report for Xcode permission prompt behavior
   - Metrics comparison (direct mode vs broker mode process churn)
 - **Acceptance Criteria:**
-  - [ ] Sequential short-lived clients reuse one broker-owned upstream bridge process
-  - [ ] Concurrent client tool calls remain stable under load
+  - [x] Sequential short-lived clients reuse one broker-owned upstream bridge process
+  - [x] Concurrent client tool calls remain stable under load
   - [ ] Manual test confirms no extra Xcode prompt while broker stays running
-  - [ ] Regression suite passes with broker mode enabled
+  - [x] Regression suite passes with broker mode enabled
 
 ---