fix(cli,gateway): complete max_tokens propagation — CLI path + env var override

Previous commit only covered the gateway runtime path. This adds:
- CLI __init__: read max_tokens from model config with HERMES_MAX_TOKENS env override
- CLI AIAgent() calls (interactive + background): pass max_tokens
- Gateway _resolve_runtime_agent_kwargs: add HERMES_MAX_TOKENS env override

All three code paths (CLI, gateway runtime, session override) now
consistently propagate max_tokens to AIAgent.

cli.py

@@ -3194,6 +3194,18 @@ class HermesCLI:
         _config_model = (_model_config.get("default") or _model_config.get("model") or "") if isinstance(_model_config, dict) else (_model_config or "")
         _DEFAULT_CONFIG_MODEL = ""
         self.model = model or _config_model or _DEFAULT_CONFIG_MODEL
+        # Read max_tokens from config (env var override: HERMES_MAX_TOKENS)
+        _env_mt = os.environ.get("HERMES_MAX_TOKENS")
+        if _env_mt:
+            try:
+                self.max_tokens = int(_env_mt)
+            except (ValueError, TypeError):
+                self.max_tokens = None
+        elif isinstance(_model_config, dict):
+            _mt = _model_config.get("max_tokens")
+            self.max_tokens = _mt if isinstance(_mt, int) else None
+        else:
+            self.max_tokens = None
         # Auto-detect model from local server if still on default
         if self.model == _DEFAULT_CONFIG_MODEL:
             _base_url = (_model_config.get("base_url") or "") if isinstance(_model_config, dict) else ""
@@ -5168,6 +5180,7 @@ class HermesCLI:
                 acp_command=runtime.get("command"),
                 acp_args=runtime.get("args"),
                 credential_pool=runtime.get("credential_pool"),
+                max_tokens=self.max_tokens,
                 max_iterations=self.max_turns,
                 enabled_toolsets=self.enabled_toolsets,
                 disabled_toolsets=self.disabled_toolsets,
@@ -9284,6 +9297,7 @@ class HermesCLI:
                     api_mode=turn_route["runtime"].get("api_mode"),
                     acp_command=turn_route["runtime"].get("command"),
                     acp_args=turn_route["runtime"].get("args"),
+                    max_tokens=turn_route["runtime"].get("max_tokens"),
                     max_iterations=self.max_turns,
                     enabled_toolsets=self.enabled_toolsets,
                     quiet_mode=True,

gateway/run.py

@@ -1203,7 +1203,13 @@ def _resolve_runtime_agent_kwargs() -> dict:
 
     model_cfg = _get_model_config()
     max_tokens = None
-    if isinstance(model_cfg, dict):
+    _env_mt = os.environ.get("HERMES_MAX_TOKENS")
+    if _env_mt:
+        try:
+            max_tokens = int(_env_mt)
+        except (ValueError, TypeError):
+            max_tokens = None
+    elif isinstance(model_cfg, dict):
         mt = model_cfg.get("max_tokens")
         if isinstance(mt, int):
             max_tokens = mt