fix(mcp): correct stale Python SDK path references in known-issues.md (KI-007)

The KI-007 entry referenced `molecule_sdk_python/molecule_agent/client.py`
— the old monorepo path prefix. In the standalone `molecule-sdk-python`
repo the package is at `molecule_agent/client.py` directly. Updated all
references to clarify the standalone repo context.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

.gitea/workflows/gitea-merge-queue.yml

@@ -0,0 +1,40 @@
+name: gitea-merge-queue
+
+on:
+  schedule:
+    - cron: '*/5 * * * *'
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+concurrency:
+  group: gitea-merge-queue-${{ github.repository }}
+  cancel-in-progress: false
+
+jobs:
+  queue:
+    runs-on: ubuntu-latest
+    timeout-minutes: 5
+    steps:
+      - name: Check out queue script from main
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          ref: ${{ github.event.repository.default_branch }}
+
+      - name: Process one queued PR
+        env:
+          GITEA_TOKEN: ${{ secrets.AUTO_SYNC_TOKEN }}
+          GITEA_HOST: git.moleculesai.app
+          REPO: ${{ github.repository }}
+          WATCH_BRANCH: ${{ github.event.repository.default_branch }}
+          QUEUE_LABEL: merge-queue
+          HOLD_LABEL: merge-queue-hold
+          UPDATE_STYLE: merge
+          # Context names: CI workflow is "CI", job is "test" → "CI / test".
+          # The CI workflow (.gitea/workflows/ci.yml) triggers on pull_request events,
+          # so Gitea posts statuses with the "(pull_request)" suffix.
+          # The queue script also checks the SOP gate: sop-checklist / all-items-acked (pull_request).
+          REQUIRED_CONTEXTS: >-
+            CI / test (pull_request), sop-checklist / all-items-acked (pull_request)
+        run: python3 .gitea/scripts/gitea-merge-queue.py

known-issues.md

@@ -188,13 +188,17 @@ The MCP server's remote-agent tools (`list_remote_agents`, `get_remote_agent_sta
 `check_remote_agent_freshness`, `get_remote_agent_setup_command`) are **read-only
 queries** — they do not drive any background heartbeat loop. The actual
 `run_heartbeat_loop()` that sends heartbeats from a remote agent lives in the
-Python SDK (`molecule_sdk_python/molecule_agent/client.py`).
+Python SDK's `molecule_agent/client.py` (standalone `molecule-sdk-python` repo).
 
 The heartbeat cleanup issue (heartbeat loop continues after the controlling MCP
-client disconnects) is tracked as **SDK KI-009** in `molecule-sdk-python/known-issues.md`.
+client disconnects) is tracked as **SDK KI-009** in the `molecule-sdk-python`
+repo's `known-issues.md`. The Python SDK lives in the `molecule-sdk-python`
+standalone repo; the `RemoteAgentClient` class is at `molecule_agent/client.py`
+(not the old monorepo path `sdk/python/molecule_agent/`).
 
 ### Suggested fix (SDK side)
 Expose a `stop_event` parameter or `stop()` method on `RemoteAgentClient` so the
 callers (MCP client, shell wrapper) can signal the loop to exit cleanly. The
 Python SDK's `run_heartbeat_loop()` should check `threading.Event` or accept a
-`stop_on: asyncio.Event` argument. See `molecule-sdk-python/known-issues.md`.
+`stop_on: asyncio.Event` argument. See the `molecule-sdk-python` repo's
+`known-issues.md` (KI-009 there).