diff --git a/CHANGELOG.md b/CHANGELOG.md
index 4ad07a7..1a1a631 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,6 +9,14 @@ All notable changes to RigForge are documented here. The format is based on
 
 ### Added
 
+- **The enriched feed mirrors the last control outcome (#346).** Pithead's host runner caps its
+  synchronous `/status` poll after a worker-apply, so a slow auto-rollback outruns it and the change
+  history row froze at `accepted`. The sister API's `rigforge` block now carries `control` —
+  `{change_id, status, reason}` read from the control status file — so the next routine poll of the
+  open read feed catches the terminal outcome without a new authenticated dial to the control port.
+  A rig that never took a control change, or an unreadable/malformed status file, serves
+  `control: null`; the mirror never breaks the feed.
+
 - **`doctor` and `apply` now notice when the rig isn't mining at all (#343).** With `pools[0].url`
   pointed at an unresolvable host, XMRig loops on DNS errors forever — and `doctor` still said "all
   critical checks passed" while `apply` said "Applied config and restarted". Both now ask the miner
diff --git a/docs/operations.md b/docs/operations.md
index ce51634..ec34a5f 100644
--- a/docs/operations.md
+++ b/docs/operations.md
@@ -534,8 +534,9 @@ second read-only HTTP endpoint (default `:8081`, keys `api_port`/`api_bind`) wit
 `:8080` XMRig API has **plus** the data only RigForge knows: applied tune knobs and the last tune
 run (`/tune`), hashrate-per-watt from RAPL, the doctor probes — HugePages, MSR state, governor,
 RAM channels/speeds, memory-profile and SMT state, throttling — the watchdog's armed/thermal state,
-the effective **writable** config (`config`, pool secrets masked; #253), and a config revision +
-last-change provenance (`config_meta`; #254) — as JSON (`/health`, or nested under
+the effective **writable** config (`config`, pool secrets masked; #253), a config revision +
+last-change provenance (`config_meta`; #254), and the last control-path outcome (`control`,
+`{change_id, status, reason}` or `null`; #346) — as JSON (`/health`, or nested under
 `rigforge` in `/1/summary` and `/2/summary`). It follows XMRig's own architecture: one tiny
 persistent server (python3 stdlib, ~10 MB idle) ships pre-computed bytes, so a request costs
 microseconds and cannot touch mining performance; a systemd timer recomputes the state every 15
diff --git a/docs/pithead-integration.md b/docs/pithead-integration.md
index a097a40..70c028b 100644
--- a/docs/pithead-integration.md
+++ b/docs/pithead-integration.md
@@ -82,7 +82,10 @@ pithead#235):
   config — exactly the control-path allowlist, pool secrets masked; see the prefill note in §3), and
   `config_meta` (`{revision, changed_at, source, last_change_id}` — `revision` is a content hash of the
   writable config that changes iff that config changes, so a poller can detect a change made directly
-  on the rig; `source` is `control`/`local`/`restore`; see §3).
+  on the rig; `source` is `control`/`local`/`restore`; see §3), and `control` (the last control-path
+  outcome, `{change_id, status, reason}` mirrored from the control status file, so a poller that
+  missed a slow rollback catches the terminal outcome here without dialing the control port; `null`
+  when the rig has never taken a control change or the status file is unreadable).
 - `GET /health` and `GET /tune` — the `rigforge.health` / `rigforge.tune` objects bare.
 - When XMRig's own API is unreachable the response is still `200` with
   `"rigforge": {..., "xmrig_api": "unreachable"}` — a down miner is exactly when the health data
diff --git a/rigforge.sh b/rigforge.sh
index e01413b..cebb21c 100755
--- a/rigforge.sh
+++ b/rigforge.sh
@@ -4749,8 +4749,17 @@ _api_config_meta_json() {
     fi
 }
 
+# #346: the last control outcome mirrored into the feed as `rigforge.control`. Pithead's host runner
+# caps its synchronous /status poll after a worker-apply, so a slow rollback outruns it; rather than a
+# new authenticated dial to the control port, the rig mirrors {change_id, status, reason} from the
+# status.json _control_status already writes 644, and the next routine poll catches up. Missing,
+# unreadable, or malformed status.json -> null — the mirror must never break the feed.
+_api_control_json() {
+    jq -c '{change_id, status, reason}' "${RIGFORGE_CONTROL_STATE:-/var/lib/rigforge-control}/status.json" 2>/dev/null || echo null
+}
+
 _api_rigforge_block() { # <hashrate|"">
-    jq -n --arg v "$(cat "$SCRIPT_DIR/VERSION" 2>/dev/null || echo unknown)" --arg xv "$XMRIG_VERSION" --arg xc "$XMRIG_COMMIT" --argjson tune "$(_api_tune_json)" --argjson power "$(_api_power_json "$1")" --argjson health "$(_health_json)" --argjson watchdog "$(_watchdog_json)" --argjson config "$(_api_config_json)" --argjson config_meta "$(_api_config_meta_json)" '{version: $v, xmrig_version: $xv, xmrig_commit: $xc, tune: $tune, power: $power, health: $health, watchdog: $watchdog, config: $config, config_meta: $config_meta}'
+    jq -n --arg v "$(cat "$SCRIPT_DIR/VERSION" 2>/dev/null || echo unknown)" --arg xv "$XMRIG_VERSION" --arg xc "$XMRIG_COMMIT" --argjson tune "$(_api_tune_json)" --argjson power "$(_api_power_json "$1")" --argjson health "$(_health_json)" --argjson watchdog "$(_watchdog_json)" --argjson config "$(_api_config_json)" --argjson config_meta "$(_api_config_meta_json)" --argjson control "$(_api_control_json)" '{version: $v, xmrig_version: $xv, xmrig_commit: $xc, tune: $tune, power: $power, health: $health, watchdog: $watchdog, config: $config, config_meta: $config_meta, control: $control}'
 }
 
 # Produce the sister API's response bodies: compute once, write atomically (tmp + rename, the
diff --git a/tests/run.sh b/tests/run.sh
index 4fc5b18..9d5b977 100644
--- a/tests/run.sh
+++ b/tests/run.sh
@@ -6094,6 +6094,20 @@ printf '{broken' >"$APIQ/home/worker/tune-overrides.json"
 run_refresh
 assert_eq "corrupt tune-overrides -> applied null, not a crash" "$(jq -r '.applied' "$APIQ/data/tune.json")" "null"
 rm -f "$APIQ/home/worker/tune-overrides.json"
+# #346: the last control outcome mirrored into the feed as rigforge.control — pithead's poller catches
+# a late terminal outcome on the open read feed instead of a new authenticated dial to the control port.
+# A realistic full status.json (the _control_status shape): the mirror picks exactly the three keys.
+CTL346="$(mktemp -d "$SANDBOX/ctl346.XXXXXX")"
+printf '%s' '{"status":"rolled_back","change_id":"abc0123456789def","source":"control","applied_at":"2026-01-01T00:00:00Z","changed_keys":["DONATION"],"reason":"miner did not return to a live hashrate; rolled back and live","backup":"/b","warnings":[]}' >"$CTL346/status.json"
+run_refresh "RIGFORGE_CONTROL_STATE=$CTL346"
+assert_eq "control mirror: exactly {change_id, status, reason} on the feed (#346)" "$(jq -cS '.rigforge.control' "$APIQ/data/summary.json")" '{"change_id":"abc0123456789def","reason":"miner did not return to a live hashrate; rolled back and live","status":"rolled_back"}'
+# No status.json (a rig that never took a control change, or control disabled) -> null, feed intact.
+run_refresh "RIGFORGE_CONTROL_STATE=$CTL346/absent"
+assert_eq "control mirror: no status.json -> null (#346)" "$(jq -c '.rigforge.control' "$APIQ/data/summary.json")" "null"
+# Malformed status.json -> null, and the refresh still writes the feed.
+printf '{broken' >"$CTL346/status.json"
+run_refresh "RIGFORGE_CONTROL_STATE=$CTL346"
+assert_eq "control mirror: malformed status.json -> null, refresh survives (#346)" "$(jq -c '.rigforge.control' "$APIQ/data/summary.json")" "null"
 # #276 (item 5): each `printf | jq ... && mv` (rigforge.sh:4233-4235) is independently atomic — a jq
 # failure on ONE file must not corrupt or block the others. Break _api_rigforge_block so specifically
 # health.json's own extraction (`.health + {watchdog: .watchdog}`) fails (health is a string, not an