Target railiance01 explicitly for founder inventory execution

Assistant: codex
Assistant-Model: gpt-6.1-sol
Assistant-Session: 01a1abf1-7e95-7141-87c9-ee78332bf790

docs/evidence/2026-11-05-vergabe-founder-inventory-attempt.json

@@ -0,0 +1,13 @@
+{
+  "schema": "railiance.vergabe-pilot-native-inventory.v1",
+  "observed_at": "2026-11-05T22:05:14.439011+00:00",
+  "evidence_kind": "user_reported_terminal_output",
+  "workplan": "RAPPS-WP-0014",
+  "task": "RAPPS-WP-0014-T03",
+  "runtime_mutations": false,
+  "native_inventory_complete": false,
+  "cluster_read": "unavailable_or_invalid_response",
+  "execution_host": "not_reported",
+  "cluster_identity_or_outage_established": false,
+  "followup": "Use the reviewed workstation transport to connect explicitly to railiance01; return the remote inventory JSON."
+}

docs/proposals/2026-11-05-vergabe-recovery-preflight/README.md

@@ -43,8 +43,26 @@ path was unavailable in the invoking terminal; its host was not reported.
 The founder does not need an executor installation to perform their own
 reviewed read-only check.
 
-From a synced `railiance-apps` checkout in the founder's own terminal on
-**railiance01**, run:
+The subsequent direct invocation returned `cluster_read:
+unavailable_or_invalid_response`, without establishing a cluster identity. Its
+execution host is unconfirmed. Use the explicit remote transport to avoid
+depending on the workstation's root PATH or kubeconfig.
+
+From the **workstation** `railiance-apps` checkout in the founder's own
+interactive terminal, run:
+
+```bash
+git pull --ff-only
+python3 tools/vergabe_pilot_inventory_remote.py --run
+```
+
+The transport validates the unchanged inventory source SHA256, connects through
+the existing `ssh railiance01` login with a TTY and runs the pinned source with
+the node's existing sudo/kubeconfig context. Source travels as one shell-quoted
+argument, preserving stdin for the sudo prompt. There is no remote checkout or
+local root Kubernetes context requirement. Default mode omits SSH execution.
+
+When already in a synced checkout **on railiance01**, the equivalent command is:
 
 ```bash
 sudo env KUBECONFIG=/etc/rancher/k3s/k3s.yaml /usr/bin/python3 tools/vergabe_pilot_inventory.py

tools/vergabe_pilot_inventory_remote.py

@@ -0,0 +1,59 @@
+#!/usr/bin/env python3
+"""Run the pinned read-only inventory through the founder's existing SSH login.
+
+Run from the workstation checkout in the founder's own interactive terminal.
+Default only reviews the transport. --run connects to railiance01 explicitly;
+SSH and sudo authenticate in that terminal. No administrative credential leaves
+the node. The public source travels as a quoted argument, preserving stdin for
+the sudo prompt and avoiding a remote checkout requirement.
+"""
+
+import argparse
+import hashlib
+import json
+import shlex
+import subprocess
+import sys
+from pathlib import Path
+
+SOURCE_SHA256 = "b610059b5d7398138062de51edab3555b973cfbb61f02fa684184852805d7507"
+SOURCE = Path(__file__).with_name("vergabe_pilot_inventory.py")
+
+
+def remote_command(source):
+    if hashlib.sha256(source.encode()).hexdigest() != SOURCE_SHA256:
+        raise ValueError("inventory_source_pin_mismatch")
+    operation = [
+        "sudo", "/usr/bin/env", "KUBECONFIG=/etc/rancher/k3s/k3s.yaml",
+        "/usr/bin/python3", "-c", source,
+    ]
+    return ["ssh", "-t", "railiance01", shlex.join(operation)]
+
+
+def main(argv=None):
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--run", action="store_true")
+    args = parser.parse_args(argv)
+    command = remote_command(SOURCE.read_text())
+    if not args.run:
+        print(json.dumps({"status": "prepared", "ssh_target": "railiance01",
+                          "inventory_source_sha256": SOURCE_SHA256,
+                          "runtime_mutations": False}))
+        return 0
+    if not sys.stdin.isatty():
+        raise ValueError("founder_interactive_terminal_required")
+    return subprocess.run(command, check=False).returncode
+
+
+if __name__ == "__main__":
+    try:
+        raise SystemExit(main())
+    except ValueError as error:
+        reason = str(error)
+        if reason not in {"inventory_source_pin_mismatch", "founder_interactive_terminal_required"}:
+            reason = "invalid_transport_input"
+        print(json.dumps({"status": "launcher_failed", "failure": reason}))
+        raise SystemExit(1)
+    except OSError:
+        print(json.dumps({"status": "launcher_failed", "failure": "source_or_ssh_unavailable"}))
+        raise SystemExit(1)

workplans/RAPPS-WP-0014-vergabe-invited-pilot.md

@@ -196,7 +196,7 @@ id: RAPPS-WP-0014-T03
 status: wait
 needs_human: true
 decision_id: d06307e5-2737-461a-a2f4-a12c3b492a9a
-intervention_note: "Two-account functional walkthrough through Zuschlag/Nachlese is user-confirmed with UI follow-up. Next: founder runs the reviewed read-only inventory on railiance01 using the existing supervisor-execution lane. Gexec launcher stopped before submission because its binary was missing in the invoking terminal; host is unconfirmed. Agent privileges stay unchanged. Populated restart/off-host recovery, isolated restore, rollback, access evidence and ownership remain open."
+intervention_note: "Functional walkthrough is confirmed with UI follow-up. Latest founder inventory returned cluster_read unavailable_or_invalid_response; invoking host is unconfirmed, so no cluster fault is inferred. Next: run tools/vergabe_pilot_inventory_remote.py --run from the founder workstation terminal to target railiance01 explicitly. Agent privileges stay unchanged. Native inventory, populated recovery/access/rollback and ownership remain open."
 priority: high
 assignee: the-custodian
 blocking_reason: "Two-account functional walkthrough is confirmed with UI follow-up. Await attended native inventory, populated restart, coherent PostgreSQL/media/issue-state off-host recovery, isolated restored workflow, rollback/access evidence and operating ownership. September 25 proves database-only restore and restart."
@@ -356,6 +356,20 @@ The unnecessary custom founder launcher is retired from current source; its
 earlier preparation and failed attempt remain historical evidence. The pinned
 inventory helper and original agent gexec action remain unchanged.
 
+2026-11-05 direct inventory return: user-reported JSON at 22:05:14Z has
+`native_inventory_complete: false` and `cluster_read:
+unavailable_or_invalid_response`. Host identity is unconfirmed; this establishes
+neither a valid node inventory nor a cluster outage. Receipt:
+`docs/evidence/2026-11-05-vergabe-founder-inventory-attempt.json`.
+
+Prepared `tools/vergabe_pilot_inventory_remote.py` for the founder's workstation
+terminal. It verifies the unchanged inventory source pin and connects explicitly
+through SSH to railiance01, running that source as a quoted argument under the
+node's existing sudo context. Stdin remains available for normal authentication;
+no remote checkout, local admin kubeconfig or agent privilege grant is needed.
+Default mode only reviews transport. The agent did not execute the founder's
+SSH/sudo path. Native inventory and all operational recovery evidence remain open.
+
 
 ## Validate synthetic sizing against the exact release and shared contention