chore: let admins interrupt runaway agent runs (#195)

* fix: let admins interrupt runaway agents (#1730)

Add a workspace-wide admin control that cancels every active run while preserving thread history.

(cherry picked from commit 09eaf94c3e612db969daa000b909805963fe1de9)

* chore: reconcile triage ledger

Signed-off-by: Adam Moussa <adam@seahavenind.com>

* fix: change ui import from refactor

Signed-off-by: Adam Moussa <adam@seahavenind.com>

---------

Signed-off-by: Adam Moussa <adam@seahavenind.com>
Co-authored-by: Johannes du Plessis <johannes@langchain.dev>

agent/dashboard/routes.py

@@ -152,6 +152,7 @@ from .team_settings import (
 from .thread_api import (
     ThreadMessageBody,
     ThreadResolveBody,
+    admin_cancel_dashboard_thread,
     cancel_dashboard_thread,
     delete_dashboard_thread,
     get_dashboard_thread,
@@ -1684,6 +1685,14 @@ async def api_cancel_thread(
     return await cancel_dashboard_thread(thread_id, session["sub"], email=session.get("email"))
 
 
+@router.post("/admin/threads/{thread_id}/cancel")
+async def admin_cancel_thread(
+    thread_id: str,
+    _admin: dict[str, Any] = _ADMIN_DEP,
+) -> dict[str, Any]:
+    return await admin_cancel_dashboard_thread(thread_id)
+
+
 @router.delete("/threads/{thread_id}")
 async def api_delete_thread(
     thread_id: str,

agent/dashboard/thread_api.py

@@ -1401,6 +1401,32 @@ async def cancel_dashboard_thread(
     )
 
 
+async def admin_cancel_dashboard_thread(thread_id: str) -> dict[str, Any]:
+    client = langgraph_client()
+    try:
+        thread = await client.threads.get(thread_id)
+    except Exception as exc:  # noqa: BLE001
+        raise HTTPException(404, "thread not found") from exc
+
+    metadata = thread.get("metadata") if isinstance(thread.get("metadata"), dict) else {}
+    try:
+        await client.runs.cancel_many(thread_id=thread_id, status="all", action="interrupt")
+    except Exception as exc:  # noqa: BLE001
+        logger.exception("Failed to cancel active runs for thread %s", thread_id)
+        raise HTTPException(502, "failed to request thread cancellation") from exc
+
+    await client.threads.update(
+        thread_id=thread_id,
+        metadata={"latest_run_status": "interrupted", "updated_at_ms": _now_ms()},
+    )
+    updated_thread = await client.threads.get(thread_id)
+    return _thread_summary(
+        updated_thread
+        if isinstance(updated_thread, dict)
+        else {"thread_id": thread_id, "metadata": metadata}
+    )
+
+
 async def delete_dashboard_thread(thread_id: str, login: str, *, email: str | None = None) -> None:
     client = langgraph_client()
     try:

docs/upstream-sync/triage.jsonl

@@ -96,7 +96,7 @@
 {"sha": "ddbe457b", "pr": 1727, "subject": "fix: restore GPT-5.5 as default model (#1727)", "disposition": "deferred", "reason": "restores GPT-5.5 default in options/team_settings; fork picker is Bedrock/Fireworks-only — rides the #1708 OpenAI-models product decision (27b0ddeb)", "branch": "model-picker", "local_sha": null, "updated": "2026-07-16T18:46:38Z"}
 {"sha": "1ea03a43", "pr": 1729, "subject": "feat: include Cargo in sandbox image (#1729)", "disposition": "landed", "reason": "2-line Dockerfile add (Cargo); adopt with #1728", "branch": "chore/upstream-easy-picks", "local_sha": null, "updated": "2026-07-16T19:15:42Z"}
 {"sha": "5136079d", "pr": 1725, "subject": "chore: disable todos for GPT-5.6 Sol (#1725)", "disposition": "wont-merge", "reason": "superseded — #1733 (136d28e6) rewrites the same todo-exclusion block to a global default-off with env opt-in; per-model GPT-5.6 Sol list moot (fork picker has no OpenAI models)", "branch": "", "local_sha": null, "updated": "2026-07-16T18:46:21Z"}
-{"sha": "09eaf94c", "pr": 1730, "subject": "fix: let admins interrupt runaway agents (#1730)", "disposition": "deferred", "reason": "admin interrupt for runaway agents (dashboard route + UI); useful ops control; UI half on post-reorg ui/src/features — remap to ui/src/components/agents", "branch": "dashboard-ui", "local_sha": null, "updated": "2026-07-16T18:47:07Z"}
+{"sha": "09eaf94c", "pr": 1730, "subject": "fix: let admins interrupt runaway agents (#1730)", "disposition": "landed", "reason": "admin interrupt for runaway agents (dashboard route + UI); useful ops control; UI half on post-reorg ui/src/features — remap to ui/src/components/agents", "branch": "feat/admin-thread-interrupt", "local_sha": null, "updated": "2026-07-16T19:36:07Z"}
 {"sha": "30832d29", "pr": 1731, "subject": "fix: preserve OpenAI Responses tool history (#1731)", "disposition": "deferred", "reason": "deletes SanitizeOpenAIResponsesMiddleware in favor of replay-history preservation in utils/model.py; supersedes deferred #1718 (35659177) — triage the pair together against fork-diverged middleware + model.py", "branch": "openai-sanitize", "local_sha": null, "updated": "2026-07-16T18:46:38Z"}
 {"sha": "1ea0e600", "pr": 1736, "subject": "fix: bind cached GitHub tokens to users (#1736)", "disposition": "deferred", "reason": "FLAG-HUMAN: security fix — binds per-thread cached GitHub tokens to a user principal (closes cross-user token-reuse leak). Fork has github_token.py but not webhooks/common.py — hand-map; auth surface: GPT-4.1 cross-review + /sh-security-review on landing. Priority pick.", "branch": "github-token-binding", "local_sha": null, "updated": "2026-07-16T18:46:37Z"}
 {"sha": "3fcb27ce", "pr": 1737, "subject": "fix: bound reviewer diff fetching (#1737)", "disposition": "deferred", "reason": "bounds reviewer diff fetching + new fetch_review_diff tool; reviewer.py fork-diverged — reconcile like #1713 (71e3b818)", "branch": "reviewer-misc", "local_sha": null, "updated": "2026-07-16T18:46:52Z"}

docs/upstream-sync/triage.md

@@ -67,6 +67,7 @@ Rows key on the **upstream SHA** (stable across local cherry-picks). Deferred ro
 | `22e024cb` | #1704 | fix: link issue PRs and prompt repo conventions (#1704) | Landed | issue/PR linking + repo-convention prompt; clean but prompt-conflict risk vs #113 | chore/upstream-easy-picks |
 | `129ddcf9` | #1728 | chore: include ripgrep in sandbox image (#1728) | Landed | 1-line Dockerfile add (ripgrep); dev image lacks it; trivial pick | chore/upstream-easy-picks |
 | `1ea03a43` | #1729 | feat: include Cargo in sandbox image (#1729) | Landed | 2-line Dockerfile add (Cargo); adopt with #1728 | chore/upstream-easy-picks |
+| `09eaf94c` | #1730 | fix: let admins interrupt runaway agents (#1730) | Landed | admin interrupt for runaway agents (dashboard route + UI); useful ops control; UI half on post-reorg ui/src/features — remap to ui/src/components/agents | feat/admin-thread-interrupt |
 | `c69459ad` | #1751 | fix: prefer LangSmith tools for trace links (#1751) | Landed | 1-line prompt: prefer LangSmith tools for trace links; trivial but edits fork-customized prompt.py | chore/upstream-easy-picks |
 | `5cb2e2bb` | #1750 | fix: add trace link to error banner (#1750) | Landed | adds trace link to error banner (13 lines); remap AgentThreadView.tsx path (fork: ui/src/components/agents/) | chore/upstream-easy-picks |
 | `c3292d82` | #1611 | bake sfw binary into sandbox image | Won't merge | already in dev |  |
@@ -114,7 +115,6 @@ Rows key on the **upstream SHA** (stable across local cherry-picks). Deferred ro
 | `8356eb34` | #1726 | refactor: organize repository by domain (#1726) | Deferred | FLAG-HUMAN: 298-file structural reorg (tests/<domain>/, ui/src/features/); chain head — every later upstream commit is written against this layout. Fork policy defers structural refactors (CLAUDE.md); adopting is a dedicated merge exercise. Until then, later picks need path remapping. | domain-reorg |
 | `83abea26` | #1724 | fix: accept natural-language Slack plan approvals (#1724) | Deferred | natural-language Slack plan approvals; touches fork-diverged plan-mode + Slack webhook stack (#130); post-reorg test paths need remap | plan-approval |
 | `ddbe457b` | #1727 | fix: restore GPT-5.5 as default model (#1727) | Deferred | restores GPT-5.5 default in options/team_settings; fork picker is Bedrock/Fireworks-only — rides the #1708 OpenAI-models product decision (27b0ddeb) | model-picker |
-| `09eaf94c` | #1730 | fix: let admins interrupt runaway agents (#1730) | Deferred | admin interrupt for runaway agents (dashboard route + UI); useful ops control; UI half on post-reorg ui/src/features — remap to ui/src/components/agents | dashboard-ui |
 | `30832d29` | #1731 | fix: preserve OpenAI Responses tool history (#1731) | Deferred | deletes SanitizeOpenAIResponsesMiddleware in favor of replay-history preservation in utils/model.py; supersedes deferred #1718 (35659177) — triage the pair together against fork-diverged middleware + model.py | openai-sanitize |
 | `1ea0e600` | #1736 | fix: bind cached GitHub tokens to users (#1736) | Deferred | FLAG-HUMAN: security fix — binds per-thread cached GitHub tokens to a user principal (closes cross-user token-reuse leak). Fork has github_token.py but not webhooks/common.py — hand-map; auth surface: GPT-4.1 cross-review + /sh-security-review on landing. Priority pick. | github-token-binding |
 | `3fcb27ce` | #1737 | fix: bound reviewer diff fetching (#1737) | Deferred | bounds reviewer diff fetching + new fetch_review_diff tool; reviewer.py fork-diverged — reconcile like #1713 (71e3b818) | reviewer-misc |

tests/test_dashboard_thread_api.py

@@ -1508,3 +1508,94 @@ async def test_options_gates_stale_fable_default_when_disabled() -> None:
     assert payload["default_agent_subagent_model"] != _FABLE
     assert payload["default_agent_model"] in model_ids
     assert payload["default_agent_subagent_model"] in model_ids
+
+
+async def test_admin_cancel_dashboard_thread_interrupts_all_active_runs(monkeypatch) -> None:
+    calls: list[tuple[str, dict[str, object]]] = []
+    thread = {
+        "thread_id": "thread-1",
+        "status": "busy",
+        "metadata": {
+            "title": "Runaway thread",
+            "latest_run_status": "running",
+            "updated_at_ms": 1,
+        },
+    }
+
+    class FakeThreads:
+        async def get(self, thread_id: str) -> dict[str, object]:
+            assert thread_id == "thread-1"
+            return thread
+
+        async def update(self, **kwargs: object) -> None:
+            calls.append(("update", kwargs))
+            metadata = kwargs["metadata"]
+            assert isinstance(metadata, dict)
+            thread["metadata"].update(metadata)
+
+    class FakeRuns:
+        async def cancel_many(self, **kwargs: object) -> None:
+            calls.append(("cancel_many", kwargs))
+
+    class FakeClient:
+        threads = FakeThreads()
+        runs = FakeRuns()
+
+    monkeypatch.setattr(thread_api, "langgraph_client", lambda: FakeClient())
+
+    result = await thread_api.admin_cancel_dashboard_thread("thread-1")
+
+    assert calls[0] == (
+        "cancel_many",
+        {"thread_id": "thread-1", "status": "all", "action": "interrupt"},
+    )
+    assert calls[1][0] == "update"
+    assert thread["metadata"]["latest_run_status"] == "interrupted"
+    assert result["id"] == "thread-1"
+
+
+async def test_admin_cancel_dashboard_thread_does_not_update_on_cancel_failure(monkeypatch) -> None:
+    updated = False
+
+    class FakeThreads:
+        async def get(self, thread_id: str) -> dict[str, object]:
+            return {"thread_id": thread_id, "status": "busy", "metadata": {}}
+
+        async def update(self, **kwargs: object) -> None:
+            nonlocal updated
+            updated = True
+
+    class FakeRuns:
+        async def cancel_many(self, **kwargs: object) -> None:
+            raise RuntimeError("runtime unavailable")
+
+    class FakeClient:
+        threads = FakeThreads()
+        runs = FakeRuns()
+
+    monkeypatch.setattr(thread_api, "langgraph_client", lambda: FakeClient())
+
+    with pytest.raises(HTTPException) as exc_info:
+        await thread_api.admin_cancel_dashboard_thread("thread-1")
+
+    assert exc_info.value.status_code == 502
+    assert updated is False
+
+
+async def test_admin_cancel_thread_route_delegates_without_owner_identity(monkeypatch) -> None:
+    cancel = AsyncMock(return_value={"id": "thread-1", "status": "interrupted"})
+    monkeypatch.setattr(routes, "admin_cancel_dashboard_thread", cancel)
+
+    result = await routes.admin_cancel_thread("thread-1", _admin={"sub": "admin"})
+
+    assert result == {"id": "thread-1", "status": "interrupted"}
+    cancel.assert_awaited_once_with("thread-1")
+
+
+def test_admin_cancel_thread_dependency_rejects_non_admin(monkeypatch) -> None:
+    monkeypatch.setenv("CONFIGURED_ADMINS", "admin")
+
+    with pytest.raises(HTTPException) as exc_info:
+        routes._require_admin({"sub": "not-admin", "email": "user@example.com"})
+
+    assert exc_info.value.status_code == 403

ui/src/lib/agents/api.ts

@@ -110,6 +110,7 @@ export interface WorkflowApprovalsPayload {
 export interface ThreadsPageParams {
   limit?: number
   offset?: number
+  all?: boolean
   resolved?: boolean
   viewed?: boolean
   source?: string
@@ -211,6 +212,7 @@ function buildThreadsPageQuery(params: ThreadsPageParams): string {
   const search = new URLSearchParams()
   if (params.limit != null) search.set("limit", String(params.limit))
   if (params.offset != null) search.set("offset", String(params.offset))
+  if (params.all != null) search.set("all", String(params.all))
   if (params.resolved != null) search.set("resolved", String(params.resolved))
   if (params.viewed != null) search.set("viewed", String(params.viewed))
   if (params.source) search.set("source", params.source)
@@ -293,6 +295,13 @@ export const agentsApi = {
         method: "POST",
       }
     ),
+  adminCancelThread: (threadId: string) =>
+    agentsRequest<AgentThread>(
+      `/admin/threads/${encodeURIComponent(threadId)}/cancel`,
+      {
+        method: "POST",
+      }
+    ),
   deleteThread: (threadId: string) =>
     agentsRequest<void>(`/threads/${encodeURIComponent(threadId)}`, {
       method: "DELETE",

ui/src/lib/agents/queries.ts

@@ -284,6 +284,18 @@ export function useCancelAgentThread(threadId: string) {
   })
 }
 
+export function useAdminCancelAgentThread() {
+  const queryClient = useQueryClient()
+
+  return useMutation({
+    mutationFn: (threadId: string) => agentsApi.adminCancelThread(threadId),
+    onSuccess: (thread) => {
+      queryClient.setQueryData(agentThreadKeys.detail(thread.id), thread)
+      invalidateAgentThreadLists(queryClient)
+    },
+  })
+}
+
 export function useDeleteAgentThread() {
   const queryClient = useQueryClient()
   const navigate = useNavigate()

ui/src/routes/admin.tsx

@@ -24,6 +24,10 @@ import {
 import { Skeleton } from "@/components/ui/skeleton"
 import { Switch } from "@/components/ui/switch"
 import { api } from "@/lib/api"
+import {
+  useAdminCancelAgentThread,
+  useThreadsPage,
+} from "@/lib/agents/queries"
 import { RequireLogin } from "@/lib/auth-redirect"
 import { useSession } from "@/lib/session"
 
@@ -62,6 +66,8 @@ function AdminPage() {
 
       <TriggerReviewSection />
 
+      <RunningAgentsSection />
+
       <SettingsSection title="Evals">
         <Link
           to="/admin/evals"
@@ -89,6 +95,97 @@ function AdminPage() {
   )
 }
 
+function RunningAgentsSection() {
+  const threads = useThreadsPage({
+    all: true,
+    status: "running",
+    limit: 50,
+  })
+  const cancel = useAdminCancelAgentThread()
+  const [message, setMessage] = useState<string | null>(null)
+
+  return (
+    <SettingsSection
+      title="Running agents"
+      description="Workspace-wide active threads. Killing a thread requests interruption of all pending and running runs without deleting its history."
+    >
+      <div className="flex flex-col gap-3 p-4">
+        <div className="flex items-center justify-between">
+          <span className="text-xs text-muted-foreground">
+            {threads.data?.items.length ?? 0} running
+          </span>
+          <Button
+            size="sm"
+            variant="outline"
+            onClick={() => void threads.refetch()}
+            disabled={threads.isFetching}
+          >
+            {threads.isFetching ? "Refreshing…" : "Refresh"}
+          </Button>
+        </div>
+
+        {threads.isLoading ? (
+          <Skeleton className="h-20" />
+        ) : threads.data?.items.length ? (
+          <div className="flex flex-col">
+            {threads.data.items.map((thread) => {
+              const isCancelling =
+                cancel.isPending && cancel.variables === thread.id
+              return (
+                <div
+                  key={thread.id}
+                  className="flex items-center justify-between gap-3 border-b border-border py-2 last:border-b-0"
+                >
+                  <Link
+                    to="/agents/$threadId"
+                    params={{ threadId: thread.id }}
+                    className="min-w-0 flex-1 hover:underline"
+                  >
+                    <p className="truncate text-xs font-medium text-foreground">
+                      {thread.title}
+                    </p>
+                    <p className="truncate font-mono text-[11px] text-muted-foreground">
+                      {thread.repoFullName || "no repo"} · {thread.id}
+                    </p>
+                  </Link>
+                  <Button
+                    size="sm"
+                    variant="destructive"
+                    disabled={cancel.isPending}
+                    onClick={() => {
+                      setMessage(null)
+                      cancel.mutate(thread.id, {
+                        onSuccess: () =>
+                          setMessage(`Interruption requested for ${thread.title}.`),
+                        onError: (error: Error) => setMessage(error.message),
+                      })
+                    }}
+                  >
+                    {isCancelling ? "Killing…" : "Kill"}
+                  </Button>
+                </div>
+              )
+            })}
+          </div>
+        ) : (
+          <p className="text-xs text-muted-foreground">No running agents.</p>
+        )}
+
+        {threads.error && (
+          <p className="text-xs text-destructive">{threads.error.message}</p>
+        )}
+        {message && (
+          <p
+            className={`text-xs ${cancel.isError ? "text-destructive" : "text-muted-foreground"}`}
+          >
+            {message}
+          </p>
+        )}
+      </div>
+    </SettingsSection>
+  )
+}
+
 const PR_URL_RE = /^https:\/\/github\.com\/([^/\s]+)\/([^/\s]+)\/pull\/(\d+)/
 
 function TriggerReviewSection() {