seonghobae · May 26, 2026 · May 26, 2026 · May 26, 2026 · May 26, 2026 · May 26, 2026
diff --git a/.coderabbit.yaml b/.coderabbit.yaml
@@ -1,3 +1,5 @@
-version: 2
 reviews:
   request_changes_workflow: true
+  pre_merge_checks:
+    docstrings:
+      mode: off
diff --git a/.github/workflows/bandit.yml b/.github/workflows/bandit.yml
@@ -5,6 +5,7 @@ on:
     branches: [ master ]
   pull_request:
     branches: [ master ]
+  workflow_dispatch:
 
 jobs:
   security:

diff --git a/.github/workflows/strix.yml b/.github/workflows/strix.yml
@@ -18,6 +18,7 @@ concurrency:
 permissions:
   actions: read
   contents: read
+  models: read
 
 jobs:
   strix:
@@ -80,22 +81,19 @@ jobs:
       - name: Gate Strix secrets
         id: gate
         env:
-          LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+          STRIX_GITHUB_MODELS_MODEL: ${{ secrets.STRIX_LLM || 'openai/gpt-5' }}
         run: |
-          strix_llm="github/gpt-4o"
-          llm_api_key="$(printf '%s' "$LLM_API_KEY" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')"
-          if [ -z "$llm_api_key" ]; then
-            # On protected/privileged events, fail-closed: missing secrets must
-            # not silently produce a green check.
-            if [ "$GITHUB_EVENT_NAME" = "push" ] || [ "$GITHUB_EVENT_NAME" = "schedule" ] || [ "$GITHUB_EVENT_NAME" = "pull_request_target" ]; then
-              echo '::error::Strix secrets not configured on protected or privileged event; failing closed.'
+          strix_model="$(printf '%s' "$STRIX_GITHUB_MODELS_MODEL" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')"
+          case "$strix_model" in
+            openai/gpt-5 | openai/gpt-5-* | openai/gpt-5.[0-9]* | openai/gpt-[6-9]* | openai/gpt-[1-9][0-9]* | \
+            openai/openai/gpt-5 | openai/openai/gpt-5-* | openai/openai/gpt-5.[0-9]* | openai/openai/gpt-[6-9]* | openai/openai/gpt-[1-9][0-9]*)
+              echo 'enabled=true' >> "$GITHUB_OUTPUT"
+              ;;
+            *)
+              echo '::error::STRIX_LLM must select a GitHub Models OpenAI GPT-5 or newer model, for example openai/gpt-5.'
               exit 1
-            fi
-            echo 'enabled=false' >> "$GITHUB_OUTPUT"
-            echo 'Strix secrets not configured; skipping.'
-          else
-            echo 'enabled=true' >> "$GITHUB_OUTPUT"
-          fi
+              ;;
+          esac
 
       - name: Gate GCP credentials
         id: gcp_gate
@@ -114,10 +112,11 @@ jobs:
         id: auth_gate
         if: steps.gate.outputs.enabled == 'true'
         env:
-          DEFAULT_PROVIDER: gemini
+          STRIX_GITHUB_MODELS_MODEL: ${{ secrets.STRIX_LLM || 'openai/gpt-5' }}
+          DEFAULT_PROVIDER: openai
         run: |
           . "$TRUSTED_WORKSPACE/scripts/ci/strix_model_utils.sh"
-          strix_llm="github/gpt-4o"
+          strix_llm="$(trim_whitespace "$STRIX_GITHUB_MODELS_MODEL")"
           requires_vertex_auth='false'
           model_auth_rc=0
           model_requires_vertex_auth "$strix_llm" || model_auth_rc=$?
@@ -195,7 +194,7 @@ jobs:
       - name: Mask LLM API key
         if: steps.gate.outputs.enabled == 'true' && steps.auth_gate.outputs.can_run == 'true'
         env:
-          LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+          LLM_API_KEY: ${{ github.token }}
         run: |
           # Sanitize CR/LF before masking to prevent broken ::add-mask::
           # commands and potential workflow command injection.
@@ -211,7 +210,7 @@ jobs:
       - name: Prepare LLM API key input file
         if: steps.gate.outputs.enabled == 'true' && steps.auth_gate.outputs.can_run == 'true'
         env:
-          LLM_API_KEY_SECRET: ${{ secrets.LLM_API_KEY }}
+          LLM_API_KEY_SECRET: ${{ github.token }}
         run: |
           if [ -n "$LLM_API_KEY_SECRET" ]; then
             umask 077
@@ -222,37 +221,44 @@ jobs:
 
       - name: Prepare Strix model input file
         if: steps.gate.outputs.enabled == 'true' && steps.auth_gate.outputs.can_run == 'true'
+        env:
+          STRIX_GITHUB_MODELS_MODEL: ${{ secrets.STRIX_LLM || 'openai/gpt-5' }}
         run: |
           umask 077
           strix_llm_file="$RUNNER_TEMP/strix_llm.txt"
-            printf '%s' "github/gpt-4o" > "$strix_llm_file"
+          strix_model="$(printf '%s' "$STRIX_GITHUB_MODELS_MODEL" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')"
+          case "$strix_model" in
+            openai/gpt-5 | openai/gpt-5-* | openai/gpt-5.[0-9]* | openai/gpt-[6-9]* | openai/gpt-[1-9][0-9]*)
+              printf 'openai/%s' "$strix_model" > "$strix_llm_file"
+              ;;
+            openai/openai/gpt-5 | openai/openai/gpt-5-* | openai/openai/gpt-5.[0-9]* | openai/openai/gpt-[6-9]* | openai/openai/gpt-[1-9][0-9]*)
+              printf '%s' "$strix_model" > "$strix_llm_file"
+              ;;
+            *)
+              echo '::error::STRIX_LLM must select a GitHub Models OpenAI GPT-5 or newer model, for example openai/gpt-5.'
+              exit 1
+              ;;
+          esac
           echo "STRIX_LLM_FILE=$strix_llm_file" >> "$GITHUB_ENV"
 
       - name: Prepare LLM API base input file
         if: steps.gate.outputs.enabled == 'true' && steps.auth_gate.outputs.can_run == 'true'
-        env:
-          LLM_API_BASE_SECRET: ${{ secrets.LLM_API_BASE }}
         run: |
-          if [ -n "$LLM_API_BASE_SECRET" ]; then
-            umask 077
-            llm_api_base_file="$RUNNER_TEMP/llm_api_base.txt"
-            printf '%s' "$LLM_API_BASE_SECRET" > "$llm_api_base_file"
-            echo "LLM_API_BASE_FILE=$llm_api_base_file" >> "$GITHUB_ENV"
-          fi
+          umask 077
+          llm_api_base_file="$RUNNER_TEMP/llm_api_base.txt"
+          printf '%s' "https://models.github.ai/inference" > "$llm_api_base_file"
+          echo "LLM_API_BASE_FILE=$llm_api_base_file" >> "$GITHUB_ENV"
 
       - name: Run Strix (quick)
         if: steps.gate.outputs.enabled == 'true' && steps.auth_gate.outputs.can_run == 'true'
         env:
           STRIX_LLM_FILE: ${{ env.STRIX_LLM_FILE }}
-          STRIX_LLM_DEFAULT_PROVIDER: gemini
-          GEMINI_LOCATION: GLOBAL
+          STRIX_LLM_DEFAULT_PROVIDER: openai
           LLM_API_KEY_FILE: ${{ env.LLM_API_KEY_FILE }}
           LLM_API_BASE_FILE: ${{ env.LLM_API_BASE_FILE }}
-          STRIX_GEMINI_FALLBACK_MODELS: "gemini/gemini-2.5-flash gemini/gemini-2.5-pro"
-          STRIX_VERTEX_FALLBACK_MODELS: "vertex_ai/gemini-2.5-pro vertex_ai/gemini-2.5-flash"
           STRIX_TARGET_PATH: ./
           STRIX_SOURCE_DIRS: .
-          STRIX_REASONING_EFFORT: ${{ github.event_name == 'push' && 'low' || 'minimal' }}
+          STRIX_REASONING_EFFORT: low
           STRIX_LLM_MAX_RETRIES: 1
           LLM_TIMEOUT: 90
           STRIX_MEMORY_COMPRESSOR_TIMEOUT: 10
@@ -285,5 +291,5 @@ jobs:
         with:
           name: strix-reports
           path: strix_runs/
-          if-no-files-found: warn
+          if-no-files-found: error
           retention-days: 5
diff --git a/.gitignore b/.gitignore
@@ -26,6 +26,7 @@ backend/.venv/
 venv/
 .venv/
 __pycache__/
+**/__pycache__/
 *.py[cod]
 *$py.class
 *.so

diff --git a/AGENTS.md b/AGENTS.md
@@ -10,7 +10,9 @@
   explicit `if: ${{ always() }}` upload steps when needed.
 - Prefer upgrading or removing vulnerable dependencies over downgrading patched
   packages unless compatibility evidence is recorded in the PR.
-- Strix Security Scan must use `github/gpt-4o` as the default model to bypass Vertex AI GCP credential prerequisites in PR bounds.
+- Strix Security Scan must use GitHub Models with an OpenAI GPT-5-or-newer
+  model through `models: read`; do not silently fall back to Gemini, GPT-4o, or
+  other providers when this gate is configured for PR evidence.
 
 ## PR automation and review defaults
 
@@ -64,6 +66,9 @@
   avoid single-token columns such as `id`, `title`, `status`, or `priority` on
   newly introduced objects.
 - When reviews find public/private identifier leaks, stale API fixture shapes, or recurring bug patterns, update tests, frontend mocks, E2E mocks, README examples, architecture docs, and explicitly record the anti-pattern in `AGENTS.md` so the same bug pattern does not reappear in copied examples.
+- When reviews find inert navigation/dead-space controls, either wire them to an
+  implemented workspace route/API or remove the control; do not leave
+  high-traffic drawer/sidebar entries as permanent `준비 중` copy.
 - Execution steps resulting in `Timeout`, `Fatal`, `Warn`, or `Denied` outputs are considered hard failures. Tests must run without these warnings to be considered passing.
 - DB-affecting API slices need both mocked fast tests and a real PostgreSQL
   bootstrap/smoke path before PR merge evidence is considered complete.

diff --git a/ARCHITECTURE.md b/ARCHITECTURE.md
@@ -158,8 +158,9 @@ it must never fall back to scanning trusted-base content for a modified PR path.
 Pull request scans split scoped changed files into small bounded batches before
 the timeout-driven rebalance path, so large PRs do not spend the whole required
 check budget on one oversized Strix invocation. Strix remains a required
-Medium-or-higher gate, while third-party LLM/provider warnings are tracked
-separately unless they make the scan incomplete.
+Medium-or-higher gate. The workflow uses GitHub Models through `models: read`,
+the OpenAI-compatible GitHub Models endpoint, and an OpenAI GPT-5-or-newer
+model; non-GPT-5 provider fallbacks are not valid PR evidence for this gate.
 Merge-gate governance for Strix, CodeRabbit, and required review evidence is
 documented in `docs/development/merge-gate-policy.md`.
 

diff --git a/README.md b/README.md
@@ -28,6 +28,9 @@ mail/calendar/file systems.
   open-source observability.
 - PR automation is metadata-only and uses current-head robot-review evidence plus
   required checks. Human approval is not awaited by default under repo policy.
+- Strix PR/security evidence runs through GitHub Models with an OpenAI GPT-5+
+  model and the workflow `models: read` permission; pending CodeRabbit or check
+  evidence is a wait state, not a hard blocker.
 
 ## Agentic Ontology & Auto-Organization (Planned)
 
@@ -77,6 +80,12 @@ npm run build
 npm run dev
 ```
 
+`next.config.ts` caps static generation concurrency by default
+(`NEXT_STATIC_GENERATION_MAX_CONCURRENCY=2`,
+`NEXT_STATIC_GENERATION_MIN_PAGES_PER_WORKER=50`) so constrained CI/build
+machines do not fan out excessive Node/PostCSS workers. Raise those values only
+with explicit build evidence.
+
 ## Threading proof points
 
 - Canonical thread IDs are assigned in `backend/services/threading_service.py`.

diff --git a/backend/README.md b/backend/README.md
@@ -51,11 +51,19 @@ mailbox owners.
 
 ## Warning classification
 
-Current backend test warnings are known dependency/toolchain warnings:
+Backend evidence should be collected with warnings promoted to errors whenever
+the slice is under active change:
+
+```bash
+PYTHONWARNINGS=error python3 -m pytest -q
+```
+
+Historical third-party dependency warnings should be isolated or filtered with
+an explicit rationale before merge evidence is accepted:
 
 - `starlette.formparsers` imports `multipart`, which emits a pending deprecation
   warning.
 - Compiled SWIG metadata emits `__module__` deprecation warnings during import.
 
-Threading code should not add new warnings. Treat any new warning from
-application modules as a regression.
+Threading code should not add new warnings. Treat any new application warning
+or log output containing `Timeout`, `Fatal`, `Warn`, or `Denied` as a regression.
diff --git a/backend/api/llm.py b/backend/api/llm.py
@@ -43,6 +43,8 @@ async def summarize_endpoint(request: SummarizeRequest, user_id: str | None = No
             status_code=500,
             detail="An internal server error occurred while processing the request.",
         )
+    except HTTPException:
+        raise
     except Exception:
         raise HTTPException(
             status_code=500,
@@ -69,6 +71,8 @@ async def draft_endpoint(request: DraftRequest, user_id: str | None = None, db:
             status_code=500,
             detail="An internal server error occurred while processing the request.",
         )
+    except HTTPException:
+        raise
     except Exception:
         raise HTTPException(
             status_code=500,

diff --git a/backend/tests/live/__pycache__/conftest.cpython-310-pytest-9.0.3.pyc b/backend/tests/live/__pycache__/conftest.cpython-310-pytest-9.0.3.pyc
diff --git a/backend/tests/live/__pycache__/mail_smoke_test.cpython-310-pytest-9.0.3.pyc b/backend/tests/live/__pycache__/mail_smoke_test.cpython-310-pytest-9.0.3.pyc
diff --git a/backend/tests/live/__pycache__/test_live_api_sequence.cpython-310-pytest-9.0.3.pyc b/backend/tests/live/__pycache__/test_live_api_sequence.cpython-310-pytest-9.0.3.pyc
diff --git a/backend/tests/test_llm_api.py b/backend/tests/test_llm_api.py
@@ -8,13 +8,16 @@
 
 
 class MockTenantConfig:
-    def __init__(self):
-        self.openai_api_key = "test-key"
+    def __init__(self, openai_api_key="test-key"):
+        self.openai_api_key = openai_api_key
 
 
 class MockSession:
+    def __init__(self, tenant_config=None):
+        self.tenant_config = tenant_config or MockTenantConfig()
+
     async def scalar(self, stmt):
-        return MockTenantConfig()
+        return self.tenant_config
 
 
 @pytest.fixture
@@ -74,3 +77,26 @@ def test_draft_endpoint(mock_draft, client):
     )
     assert resp.status_code == 200
     assert resp.json() == {"draft": "This is a draft reply."}
+
+
+def test_llm_endpoints_preserve_missing_key_400():
+    async def override_get_db():
+        yield MockSession(MockTenantConfig(openai_api_key=None))
+
+    app.dependency_overrides[get_db] = override_get_db
+    try:
+        with TestClient(app, headers={"X-User-Id": "testuser"}) as test_client:
+            summarize = test_client.post(
+                "/api/llm/summarize", json={"email_body": "test email"}
+            )
+            draft = test_client.post(
+                "/api/llm/draft",
+                json={"email_body": "test email", "instruction": "reply nicely"},
+            )
+    finally:
+        app.dependency_overrides.clear()
+
+    assert summarize.status_code == 400
+    assert summarize.json() == {"detail": "OpenAI API key not configured"}
+    assert draft.status_code == 400
+    assert draft.json() == {"detail": "OpenAI API key not configured"}
diff --git a/docs/operations/open-source-apm.md b/docs/operations/open-source-apm.md
@@ -2,10 +2,12 @@
 
 ## 확인된 사실 / Confirmed
 
-- `backend/main.py` currently exposes the FastAPI app and routers, but no
-  OpenTelemetry instrumentation is wired yet.
-- `docker-compose.yml` currently contains db/backend/frontend only; it does not
-  run Prometheus, Grafana, Loki, Tempo, Jaeger, or an OpenTelemetry Collector.
+- `backend/main.py` exposes Prometheus metrics and optional OpenTelemetry
+  FastAPI instrumentation.
+- `backend/requirements.txt` includes `prometheus-fastapi-instrumentator` and
+  OpenTelemetry dependencies used by the FastAPI app.
+- `docker-compose.observability.yml`, `docker-compose.apm.yml`, and
+  `docker-compose.infra.yml` document local Prometheus/APM stack entry points.
 - `docker-compose.live-e2e.yml` proves the image-based smoke path before any APM
   stack is claimed as production-ready.
 
@@ -30,8 +32,16 @@
 
 ## 도입 기준
 
-- Add instrumentation behind explicit environment variables.
-- Keep `/healthz`, `/readyz`, and `/metrics` semantics separate if those endpoints
-  are added later.
+- Keep instrumentation behind explicit environment variables where tracing can
+  export outside the process.
+- Keep `/healthz`, `/readyz`, and `/metrics` semantics separate as the stack
+  matures.
 - Do not claim APM production readiness until a live stack shows trace, metric,
   and log evidence without leaking user email content.
+
+## Remaining gaps
+
+- Add connector heartbeat, sync lag, writeback conflict, and AI action audit
+  dashboards.
+- Add log and trace redaction tests for email bodies, provider tokens, DSNs, and
+  calendar/file descriptions.
diff --git a/docs/operations/source-of-truth-and-writeback-sovereignty.md b/docs/operations/source-of-truth-and-writeback-sovereignty.md
@@ -28,6 +28,23 @@
 - GitHub self-hosted runners remain CI smoke infrastructure. They can validate
   internal connectivity, but they are not the production relay component.
 
+## Source registry contract
+
+Every writeback-capable account should resolve through a server-authoritative
+source registry before any provider mutation is attempted. The registry record
+must include an opaque `source_id`, tenant/workspace owner scope, protocol
+(`imap`, `smtp`, `pop3`, `caldav`, `carddav`, `webdav`), provider host, capability
+flags, consent state, data region, connector route, credential reference, and
+conflict metadata such as remote ids or ETags. Browser requests may reference
+`source_id`, but must never supply ownership, credential, region, or capability
+claims.
+
+Implemented slices currently cover signed task/mail provenance, dedupe/threading
+contracts, source-linked ticket tasks, and writeback intent documentation. Real
+CalDAV/WebDAV mutation, POP3 runtime sync, sent-mail reply tracking persistence,
+self-sent knowledge capture, and sender DAG/ontology action selection remain
+episode work and must preserve this registry boundary.
+
 ## Policy and audit requirements
 
 - RBAC grants are necessary but never sufficient when ABAC denies apply. Region,

diff --git a/docs/plans/2026-05-17-startup-view-responsive-mobile-menu.md b/docs/plans/2026-05-17-startup-view-responsive-mobile-menu.md
@@ -32,7 +32,7 @@
 
 - [x] **Step 1: Write failing unit tests**
 
-Update `frontend/src/app/page.test.tsx` with tests that seed `localStorage.setItem('naruon_startup_view', 'calendar')` and expect the mobile calendar region to be the active startup panel, seed `dashboard` and expect a dashboard overview region, and verify the default remains the email workspace.
+Update `frontend/src/app/page.test.tsx` with tests that seed `localStorage.setItem('naruon_startup_view', 'calendar')` and expect the mobile calendar region to be the active startup panel, seed `dashboard` and expect a dashboard overview region, and verify the default opens the dashboard workspace.
 
 - [x] **Step 2: Verify RED**