fix(run_agent): read context_length from providers/custom_providers for compression model

agent/prompt_builder.py

@@ -350,7 +350,13 @@ def _strip_yaml_frontmatter(content: str) -> str:
     ),
     "cli": (
         "You are a CLI AI Agent. Try not to use markdown but simple text "
-        "renderable inside a terminal."
+        "renderable inside a terminal. "
+        "File delivery: there is no attachment channel — the user reads your "
+        "response directly in their terminal. Do NOT emit MEDIA:/path tags "
+        "(those are only intercepted on messaging platforms like Telegram, "
+        "Discord, Slack, etc.; on the CLI they render as literal text). "
+        "When referring to a file you created or changed, just state its "
+        "absolute path in plain text; the user can open it from there."
     ),
     "sms": (
         "You are communicating via SMS. Keep responses concise and use plain text "

hermes_cli/model_normalize.py

@@ -408,6 +408,23 @@ def normalize_model_for_provider(model_input: str, target_provider: str) -> str:
             return bare
         return _normalize_for_deepseek(bare)
 
+    # --- Kimi / Moonshot: map auto-discovered short IDs (k2p6) to API-accepted
+    #     canonical names (kimi-k2.6).  The /models endpoint returns internal
+    #     short forms that fail at request time without this repair (issue #13758).
+    if provider in {"kimi-coding", "kimi-coding-cn", "moonshot"}:
+        bare = _strip_matching_provider_prefix(name, provider)
+        if "/" in bare:
+            bare = bare.split("/", 1)[1]
+        # k2p6 -> kimi-k2.6, k2p5 -> kimi-k2.5, etc.
+        _kimi_short_map = {
+            "k2p6": "kimi-k2.6",
+            "k2p5": "kimi-k2.5",
+            "k2p8": "kimi-k2.8",
+        }
+        if bare.lower() in _kimi_short_map:
+            return _kimi_short_map[bare.lower()]
+        return bare
+
     # --- Direct providers: repair matching provider prefixes only ---
     if provider in _MATCHING_PREFIX_STRIP_PROVIDERS:
         return _strip_matching_provider_prefix(name, provider)

hermes_cli/tools_config.py

@@ -80,12 +80,20 @@ def _get_effective_configurable_toolsets():
 
     Plugin toolsets are appended at the end so they appear after the
     built-in toolsets in the TUI checklist.
+    Duplicate keys (plugin toolsets whose key already exists in
+    CONFIGURABLE_TOOLSETS) are skipped so the built-in row is kept —
+    plugin-added tools are extensions of that toolset, not a second
+    top-level row (issue #13640).
     """
     result = list(CONFIGURABLE_TOOLSETS)
+    built_in_keys = {ts_key for ts_key, _, _ in CONFIGURABLE_TOOLSETS}
     try:
         from hermes_cli.plugins import discover_plugins, get_plugin_toolsets
         discover_plugins()  # idempotent — ensures plugins are loaded
-        result.extend(get_plugin_toolsets())
+        plugin_toolsets = get_plugin_toolsets()
+        # Skip plugin toolsets whose key already exists in built-ins
+        plugin_toolsets = [(k, l, d) for k, l, d in plugin_toolsets if k not in built_in_keys]
+        result.extend(plugin_toolsets)
     except Exception:
         pass
     return result

run_agent.py

@@ -2178,11 +2178,58 @@ def _check_compression_model_feasibility(self) -> None:
             aux_base_url = str(getattr(client, "base_url", ""))
             aux_api_key = str(getattr(client, "api_key", ""))
 
+            # If no explicit override, look up context_length from
+            # providers / custom_providers for the auxiliary model — the
+            # same lookup that the main model uses in __init__.
+            _aux_ctx = getattr(self, "_aux_compression_context_length_config", None)
+            if _aux_ctx is None and aux_base_url:
+                try:
+                    from hermes_cli.config import get_compatible_custom_providers
+                    _custom_providers = get_compatible_custom_providers(_agent_cfg)
+                except Exception:
+                    _custom_providers = _agent_cfg.get("custom_providers", [])
+                for _cp_entry in (_custom_providers or []):
+                    if not isinstance(_cp_entry, dict):
+                        continue
+                    _cp_url = (_cp_entry.get("base_url") or "").rstrip("/")
+                    if _cp_url and _cp_url == aux_base_url.rstrip("/"):
+                        _cp_models = _cp_entry.get("models", {})
+                        if isinstance(_cp_models, dict):
+                            _cp_model_cfg = _cp_models.get(aux_model, {})
+                            if isinstance(_cp_model_cfg, dict):
+                                _cp_ctx = _cp_model_cfg.get("context_length")
+                                if _cp_ctx is not None:
+                                    try:
+                                        _aux_ctx = int(_cp_ctx)
+                                    except (ValueError, TypeError):
+                                        pass
+                            if _aux_ctx is not None:
+                                break
+                # Also check the keyed providers schema (providers.<key>.models.<model>.context_length)
+                if _aux_ctx is None:
+                    _providers = _agent_cfg.get("providers", {})
+                    if isinstance(_providers, dict):
+                        for _pk, _pentry in _providers.items():
+                            if not isinstance(_pentry, dict):
+                                continue
+                            _purl = (_pentry.get("base_url") or "").rstrip("/")
+                            if _purl and _purl == aux_base_url.rstrip("/"):
+                                _pm = _pentry.get("models", {})
+                                if isinstance(_pm, dict):
+                                    _pm_cfg = _pm.get(aux_model, {})
+                                    if isinstance(_pm_cfg, dict):
+                                        _pm_ctx = _pm_cfg.get("context_length")
+                                        if _pm_ctx is not None:
+                                            try:
+                                                _aux_ctx = int(_pm_ctx)
+                                            except (ValueError, TypeError):
+                                                pass
+
             aux_context = get_model_context_length(
                 aux_model,
                 base_url=aux_base_url,
                 api_key=aux_api_key,
-                config_context_length=getattr(self, "_aux_compression_context_length_config", None),
+                config_context_length=_aux_ctx,
             )
 
             # Hard floor: the auxiliary compression model must have at least

tests/agent/test_prompt_builder.py

@@ -789,6 +789,24 @@ def test_platform_hints_known_platforms(self):
         assert "cron" in PLATFORM_HINTS
         assert "cli" in PLATFORM_HINTS
 
+    def test_cli_hint_does_not_suggest_media_tags(self):
+        # Regression: MEDIA:/path tags are intercepted only by messaging
+        # gateway platforms. On the CLI they render as literal text and
+        # confuse users. The CLI hint must steer the agent away from them.
+        cli_hint = PLATFORM_HINTS["cli"]
+        assert "MEDIA:" in cli_hint, (
+            "CLI hint should mention MEDIA: in order to tell the agent "
+            "NOT to use it (negative guidance)."
+        )
+        # Must contain explicit "don't" language near the MEDIA reference.
+        assert any(
+            marker in cli_hint.lower()
+            for marker in ("do not emit media", "not intercepted", "do not", "don't")
+        ), "CLI hint should explicitly discourage MEDIA: tags."
+        # Messaging hints should still advertise MEDIA: positively (sanity
+        # check that this test is calibrated correctly).
+        assert "include MEDIA:" in PLATFORM_HINTS["telegram"]
+
 
 # =========================================================================
 # Environment hints

ui-tui/src/components/appChrome.tsx

@@ -131,7 +131,10 @@ export function StatusRule({
           ) : (
             <Text color={statusColor}>{status}</Text>
           )}
-          <Text color={t.color.dim}> │ {model}</Text>
+          <Text color={t.color.dim}> │ </Text>
+          <Text color={t.color.dim} wrap="truncate-end" width={18}>
+            {model}
+          </Text>
           {ctxLabel ? <Text color={t.color.dim}> │ {ctxLabel}</Text> : null}
           {bar ? (
             <Text color={t.color.dim}>