fix: resolve remote hermes binary dynamically, validate before dispatch (#350)

Root cause: vps-agent-dispatch hardcoded /root/wizards/*/hermes/venv/bin/hermes
paths that go stale when wizards are reinstalled or moved.

- SSHEnvironment.resolve_remote_hermes_path(): dynamic resolution via
  which hermes -> ~/.local/bin/hermes -> /root/wizards/*/venv/bin/hermes
- SSHEnvironment.validate_remote_hermes(): structured check returning
  {available, path, version, error} — callers can gate dispatch on this
- topology_scan.py: replace hardcoded venv/bin paths with find-based resolution
- forge_health_check.py: default target from HERMES_WIZARDS_ROOT env var
- 6 new tests covering resolution and validation logic

Dispatch now fails loud instead of reporting OK on broken paths.

docs/bezalel/topology_scan.py

@@ -75,7 +75,7 @@ for base in ["/root/wizards/bezalel/hermes", "/root/wizards/bezalel/evennia"]:
 write("06_git_repos", "\n".join(git_repos))
 
 # 8. Python Dependencies
-venv_pip = shell("/root/wizards/bezalel/hermes/venv/bin/pip freeze 2>/dev/null | head -80")
+venv_pip = shell("$(find /root/wizards -maxdepth 4 -name pip -path '*/venv/bin/pip' 2>/dev/null | head -1) freeze 2>/dev/null | head -80")
 write("07_dependencies", f"Hermes venv packages (top 80):\n{venv_pip}")
 
 # 9. External APIs & Endpoints
@@ -115,8 +115,8 @@ mempalace = f"""MEMPALACE CONFIGURATION
 - Identity: /root/.mempalace/identity.txt
 - Config: /root/wizards/bezalel/mempalace.yaml
 - Nightly re-mine: 03:00 UTC via /root/wizards/bezalel/mempalace_nightly.sh
-- Miner binary: /root/wizards/bezalel/hermes/venv/bin/mempalace
-- Current status: {shell('/root/wizards/bezalel/hermes/venv/bin/mempalace --palace /root/wizards/bezalel/.mempalace/palace status 2>/dev/null')}
+- Miner binary: $(find /root/wizards -maxdepth 4 -name mempalace -path '*/venv/bin/mempalace' 2>/dev/null | head -1)
+- Current status: {shell('$(find /root/wizards -maxdepth 4 -name mempalace -path "*/venv/bin/mempalace" 2>/dev/null | head -1) --palace /root/wizards/bezalel/.mempalace/palace status 2>/dev/null')}
 """
 write("11_mempalace_topology", mempalace)
 

scripts/forge_health_check.py

@@ -9,8 +9,8 @@ Scans wizard environments for:
 
 Usage:
     python scripts/forge_health_check.py /root/wizards
-    python scripts/forge_health_check.py /root/wizards --json
-    python scripts/forge_health_check.py /root/wizards --fix-permissions
+    python scripts/forge_health_check.py $HERMES_WIZARDS_ROOT --json
+    python scripts/forge_health_check.py --fix-permissions
 """
 
 from __future__ import annotations
@@ -263,7 +263,7 @@ def print_report(report: HealthReport) -> None:
 
 def main(argv: list[str] | None = None) -> int:
     parser = argparse.ArgumentParser(description="Forge Health Check")
-    parser.add_argument("target", nargs="?", default="/root/wizards", help="Root path to scan")
+    parser.add_argument("target", nargs="?", default=os.environ.get("HERMES_WIZARDS_ROOT", "/root/wizards"), help="Root path to scan")
     parser.add_argument("--json", action="store_true", help="Output JSON report")
     parser.add_argument("--fix-permissions", action="store_true", help="Auto-fix file permissions")
     args = parser.parse_args(argv)

tests/test_remote_hermes_path.py

@@ -0,0 +1,175 @@
+"""Tests for #350: remote hermes path validation."""
+
+import subprocess
+from unittest.mock import MagicMock, patch, PropertyMock
+
+import pytest
+
+from tools.environments.ssh import SSHEnvironment
+
+
+class TestResolveRemoteHermesPath:
+    """Test that resolve_remote_hermes_path checks multiple locations."""
+
+    @patch("tools.environments.ssh.SSHEnvironment._establish_connection")
+    @patch("tools.environments.ssh.SSHEnvironment._detect_remote_home", return_value="/root")
+    @patch("tools.environments.ssh.SSHEnvironment._sync_skills_and_credentials")
+    def test_finds_hermes_via_which(self, mock_sync, mock_home, mock_conn):
+        """Should return path from `which hermes` when available."""
+        env = SSHEnvironment.__new__(SSHEnvironment)
+        env.host = "test-host"
+        env.user = "root"
+        env.port = 22
+        env.key_path = ""
+        env.control_socket = "/tmp/test.sock"
+
+        with patch("subprocess.run") as mock_run:
+            # First call: which hermes succeeds
+            mock_run.return_value = MagicMock(
+                stdout="/usr/local/bin/hermes\n",
+                returncode=0,
+            )
+            path = env.resolve_remote_hermes_path()
+            assert path == "/usr/local/bin/hermes"
+
+    @patch("tools.environments.ssh.SSHEnvironment._establish_connection")
+    @patch("tools.environments.ssh.SSHEnvironment._detect_remote_home", return_value="/root")
+    @patch("tools.environments.ssh.SSHEnvironment._sync_skills_and_credentials")
+    def test_falls_back_to_local_bin(self, mock_sync, mock_home, mock_conn):
+        """Should check ~/.local/bin/hermes when which fails."""
+        env = SSHEnvironment.__new__(SSHEnvironment)
+        env.host = "test-host"
+        env.user = "root"
+        env.port = 22
+        env.key_path = ""
+        env.control_socket = "/tmp/test.sock"
+
+        call_count = [0]
+
+        def mock_run_side_effect(cmd, **kwargs):
+            call_count[0] += 1
+            result = MagicMock()
+            if call_count[0] == 1:
+                # which hermes fails
+                result.stdout = ""
+                result.returncode = 1
+            elif call_count[0] == 2:
+                # ~/.local/bin/hermes exists
+                result.stdout = "/root/.local/bin/hermes\n"
+                result.returncode = 0
+            else:
+                result.stdout = ""
+                result.returncode = 1
+            return result
+
+        with patch("subprocess.run", side_effect=mock_run_side_effect):
+            path = env.resolve_remote_hermes_path()
+            assert path == "/root/.local/bin/hermes"
+
+    @patch("tools.environments.ssh.SSHEnvironment._establish_connection")
+    @patch("tools.environments.ssh.SSHEnvironment._detect_remote_home", return_value="/root")
+    @patch("tools.environments.ssh.SSHEnvironment._sync_skills_and_credentials")
+    def test_returns_empty_when_not_found(self, mock_sync, mock_home, mock_conn):
+        """Should return empty string when hermes is not found anywhere."""
+        env = SSHEnvironment.__new__(SSHEnvironment)
+        env.host = "test-host"
+        env.user = "root"
+        env.port = 22
+        env.key_path = ""
+        env.control_socket = "/tmp/test.sock"
+
+        with patch("subprocess.run") as mock_run:
+            mock_run.return_value = MagicMock(stdout="", returncode=1)
+            path = env.resolve_remote_hermes_path()
+            assert path == ""
+
+
+class TestValidateRemoteHermes:
+    """Test that validate_remote_hermes returns structured results."""
+
+    @patch("tools.environments.ssh.SSHEnvironment._establish_connection")
+    @patch("tools.environments.ssh.SSHEnvironment._detect_remote_home", return_value="/root")
+    @patch("tools.environments.ssh.SSHEnvironment._sync_skills_and_credentials")
+    def test_returns_available_when_found(self, mock_sync, mock_home, mock_conn):
+        env = SSHEnvironment.__new__(SSHEnvironment)
+        env.host = "test-host"
+        env.user = "root"
+        env.port = 22
+        env.key_path = ""
+        env.control_socket = "/tmp/test.sock"
+
+        call_count = [0]
+
+        def mock_run_side_effect(cmd, **kwargs):
+            call_count[0] += 1
+            result = MagicMock()
+            if call_count[0] == 1:
+                # which hermes
+                result.stdout = "/root/.local/bin/hermes\n"
+                result.returncode = 0
+            elif call_count[0] == 2:
+                # hermes --version
+                result.stdout = "hermes-agent 1.0.0\n"
+                result.returncode = 0
+            else:
+                result.stdout = ""
+                result.returncode = 1
+            return result
+
+        with patch("subprocess.run", side_effect=mock_run_side_effect):
+            result = env.validate_remote_hermes()
+            assert result["available"] is True
+            assert result["path"] == "/root/.local/bin/hermes"
+            assert "hermes-agent" in result["version"]
+            assert result["error"] == ""
+
+    @patch("tools.environments.ssh.SSHEnvironment._establish_connection")
+    @patch("tools.environments.ssh.SSHEnvironment._detect_remote_home", return_value="/root")
+    @patch("tools.environments.ssh.SSHEnvironment._sync_skills_and_credentials")
+    def test_returns_error_when_not_found(self, mock_sync, mock_home, mock_conn):
+        env = SSHEnvironment.__new__(SSHEnvironment)
+        env.host = "test-host"
+        env.user = "root"
+        env.port = 22
+        env.key_path = ""
+        env.control_socket = "/tmp/test.sock"
+
+        with patch("subprocess.run") as mock_run:
+            mock_run.return_value = MagicMock(stdout="", returncode=1)
+            result = env.validate_remote_hermes()
+            assert result["available"] is False
+            assert "not found" in result["error"]
+            assert result["path"] == ""
+
+    @patch("tools.environments.ssh.SSHEnvironment._establish_connection")
+    @patch("tools.environments.ssh.SSHEnvironment._detect_remote_home", return_value="/root")
+    @patch("tools.environments.ssh.SSHEnvironment._sync_skills_and_credentials")
+    def test_returns_error_when_version_fails(self, mock_sync, mock_home, mock_conn):
+        env = SSHEnvironment.__new__(SSHEnvironment)
+        env.host = "test-host"
+        env.user = "root"
+        env.port = 22
+        env.key_path = ""
+        env.control_socket = "/tmp/test.sock"
+
+        call_count = [0]
+
+        def mock_run_side_effect(cmd, **kwargs):
+            call_count[0] += 1
+            result = MagicMock()
+            if call_count[0] == 1:
+                result.stdout = "/root/.local/bin/hermes\n"
+                result.returncode = 0
+            elif call_count[0] == 2:
+                # hermes --version fails
+                result.stdout = ""
+                result.returncode = 127
+            else:
+                result.stdout = ""
+                result.returncode = 1
+            return result
+
+        with patch("subprocess.run", side_effect=mock_run_side_effect):
+            result = env.validate_remote_hermes()
+            assert result["available"] is False
+            assert "not executable" in result["error"]

tools/environments/ssh.py

@@ -154,6 +154,108 @@ class SSHEnvironment(PersistentShellMixin, BaseEnvironment):
         except Exception as e:
             logger.debug("SSH: could not sync skills/credentials: %s", e)
 
+    def resolve_remote_hermes_path(self) -> str:
+        """Resolve the hermes binary path on the remote host.
+
+        Checks in order:
+        1. `which hermes` (respects PATH, including ~/.local/bin)
+        2. ~/.local/bin/hermes (standard install location)
+        3. Common wizard paths: /root/wizards/*/hermes/venv/bin/hermes
+
+        Returns the resolved path, or empty string if not found.
+        """
+        # 1. Try which (respects PATH including ~/.local/bin if on PATH)
+        cmd = self._build_ssh_command()
+        cmd.append("which hermes 2>/dev/null")
+        try:
+            result = subprocess.run(cmd, capture_output=True, text=True, timeout=10)
+            path = result.stdout.strip()
+            if path and result.returncode == 0:
+                logger.debug("SSH: resolved hermes via which: %s", path)
+                return path
+        except Exception:
+            pass
+
+        # 2. Try ~/.local/bin/hermes (standard install symlink)
+        cmd = self._build_ssh_command()
+        cmd.append("test -x ~/.local/bin/hermes && echo ~/.local/bin/hermes")
+        try:
+            result = subprocess.run(cmd, capture_output=True, text=True, timeout=10)
+            path = result.stdout.strip()
+            if path and result.returncode == 0:
+                logger.debug("SSH: resolved hermes at ~/.local/bin/hermes")
+                return path
+        except Exception:
+            pass
+
+        # 3. Search common wizard paths
+        cmd = self._build_ssh_command()
+        cmd.append(
+            "find /root/wizards -maxdepth 4 -path '*/hermes/venv/bin/hermes' "
+            "-executable 2>/dev/null | head -1"
+        )
+        try:
+            result = subprocess.run(cmd, capture_output=True, text=True, timeout=15)
+            path = result.stdout.strip()
+            if path and result.returncode == 0:
+                logger.debug("SSH: resolved hermes at wizard path: %s", path)
+                return path
+        except Exception:
+            pass
+
+        logger.warning("SSH: could not resolve hermes binary on %s@%s", self.user, self.host)
+        return ""
+
+    def validate_remote_hermes(self) -> dict:
+        """Validate that hermes is available on the remote host.
+
+        Returns dict with:
+            available: bool
+            path: str (resolved path if found)
+            version: str (hermes --version output if available)
+            error: str (error message if not available)
+        """
+        path = self.resolve_remote_hermes_path()
+        if not path:
+            return {
+                "available": False,
+                "path": "",
+                "version": "",
+                "error": (
+                    f"Hermes binary not found on {self.user}@{self.host}. "
+                    f"Checked: which hermes, ~/.local/bin/hermes, /root/wizards/*/hermes/venv/bin/hermes. "
+                    f"Install hermes or fix the PATH."
+                ),
+            }
+
+        # Verify it's executable and get version
+        cmd = self._build_ssh_command()
+        cmd.append(f"{path} --version 2>&1 | head -1")
+        try:
+            result = subprocess.run(cmd, capture_output=True, text=True, timeout=15)
+            version = result.stdout.strip()
+            if result.returncode != 0:
+                return {
+                    "available": False,
+                    "path": path,
+                    "version": "",
+                    "error": f"Hermes at {path} exists but is not executable (exit code {result.returncode}).",
+                }
+        except subprocess.TimeoutExpired:
+            return {
+                "available": False,
+                "path": path,
+                "version": "",
+                "error": f"Hermes at {path} timed out on --version check.",
+            }
+
+        return {
+            "available": True,
+            "path": path,
+            "version": version,
+            "error": "",
+        }
+
     def execute(self, command: str, cwd: str = "", *,
                 timeout: int | None = None,
                 stdin_data: str | None = None) -> dict: