refactor: rewrite model-context to use js-yaml, add context_length to provider form (#177)

* fix: context-length API returns 200K instead of actual model context

Two bugs cause the /api/hermes/sessions/context-length endpoint to
always return DEFAULT_CONTEXT_LENGTH (200K):

1. getModelContextLength ignores config.yaml model.context_length
   The function only checks models_dev_cache.json (which doesn't
   exist in default installations) and falls back to the hardcoded
   200K default, completely ignoring the user's explicit
   model.context_length setting in config.yaml.

2. getDefaultModel regex fails when api_key/base_url come before default
   The regex /^model:\s*\n\s+default:\s*(.+)$/m assumes 'default' is
   the first child key under 'model:', but when api_key or base_url
   appear first in the YAML, the match fails. This causes
   getModelContextLength to short-circuit to DEFAULT_CONTEXT_LENGTH
   before even reaching the cache lookup.

Fix:
- Add getDefaultModelRobust() that extracts the entire model: block
  first, then searches for default: within it
- Add getConfigContextLength() that reads model.context_length from
  config.yaml as a fallback (matching hermes-agent priority)
- Update getModelContextLength() resolution order:
  1. models_dev_cache.json (existing)
  2. config.yaml model.context_length (new)
  3. DEFAULT_CONTEXT_LENGTH (existing fallback)

Closes #169

* refactor: rewrite model-context to use js-yaml, add context_length to provider form

- Replace fragile regex-based YAML parsing with js-yaml for reliable config.yaml reads
- Fix context_length resolution priority: config.yaml override > custom_providers > models_dev_cache > 200K default
- Add context_length input field when adding custom providers in ProviderFormModal
- Backend: persist context_length to custom_providers models.<model>.context_length in config.yaml
- Add i18n keys (contextLength, contextLengthPlaceholder) to all 8 locales

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: use NInputNumber instead of NInput type=number for context_length

NInput does not support type="number" in Naive UI.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: devilardis <53129661@qq.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

packages/server/src/services/hermes/model-context.ts

@@ -1,6 +1,7 @@
 import { resolve, join } from 'path'
 import { homedir } from 'os'
 import { readFileSync, existsSync, statSync } from 'fs'
+import yaml from 'js-yaml'
 
 const HERMES_BASE = resolve(homedir(), '.hermes')
 const MODELS_DEV_CACHE = resolve(HERMES_BASE, 'models_dev_cache.json')
@@ -21,6 +22,18 @@ interface ProviderEntry {
   models?: Record<string, ModelEntry>
 }
 
+// --- Config YAML helpers (js-yaml) ---
+
+function loadConfig(profileDir: string): any | null {
+  const configPath = join(profileDir, 'config.yaml')
+  if (!existsSync(configPath)) return null
+  try {
+    return yaml.load(readFileSync(configPath, 'utf-8')) as any
+  } catch {
+    return null
+  }
+}
+
 // --- In-memory cache: parsed models_dev_cache (1.7MB), invalidated by mtime ---
 
 let _cache: Record<string, ProviderEntry> | null = null
@@ -55,16 +68,59 @@ function getProfileDir(profile?: string): string {
   return existsSync(dir) ? dir : HERMES_BASE
 }
 
-function getDefaultModel(profileDir: string): string | null {
-  const configPath = join(profileDir, 'config.yaml')
-  if (!existsSync(configPath)) return null
-  try {
-    const content = readFileSync(configPath, 'utf-8')
-    const match = content.match(/^model:\s*\n\s+default:\s*(.+)$/m)
-    return match ? match[1].trim() : null
-  } catch {
-    return null
+function getDefaultModel(config: any): string | null {
+  const model = config?.model
+  if (!model || typeof model !== 'object') return null
+  return typeof model.default === 'string' ? model.default.trim() || null : null
+}
+
+function getDefaultProvider(config: any): string | null {
+  const model = config?.model
+  if (!model || typeof model !== 'object') return null
+  return typeof model.provider === 'string' ? model.provider.trim() || null : null
+}
+
+/**
+ * Read context_length from config.yaml, only as a sibling of default.
+ * e.g. model:\n  default: gpt-5.4\n  context_length: 200000
+ */
+function getConfigContextLength(config: any): number | null {
+  const model = config?.model
+  if (!model || typeof model !== 'object') return null
+  const val = model.context_length
+  if (typeof val !== 'number' || !Number.isFinite(val) || val <= 0) return null
+  return val
+}
+
+/**
+ * Lookup context_length from custom_providers in config.yaml.
+ * - "custom:xxx" → strip prefix, match by name
+ * - "custom" → match by model name
+ */
+function lookupCustomProviderContextLength(config: any, modelName: string, provider: string | null): number | null {
+  const providers: any[] = Array.isArray(config?.custom_providers) ? config.custom_providers : []
+  if (!provider || !provider.startsWith('custom')) return null
+
+  let matched: any = null
+
+  if (provider === 'custom') {
+    matched = providers.find((cp: any) => cp.model === modelName)
+  } else {
+    const suffix = provider.slice('custom:'.length)
+    matched = providers.find((cp: any) => cp.name === suffix)
   }
+
+  if (!matched) return null
+
+  const models = matched.models
+  if (!models || typeof models !== 'object') return null
+
+  const modelEntry = models[modelName]
+  if (!modelEntry || typeof modelEntry !== 'object') return null
+
+  const val = modelEntry.context_length
+  if (typeof val !== 'number' || !Number.isFinite(val) || val <= 0) return null
+  return val
 }
 
 // --- Context lookup ---
@@ -95,12 +151,33 @@ function lookupContextFromCache(modelName: string): number | null {
 
 /**
  * Get the context length for the current profile's default model.
- * Results are cached in memory (5min TTL) and invalidated by file mtime.
+ * Resolution order:
+ *   1. config.yaml model.context_length (highest priority, user override)
+ *   2. custom_providers models.<model>.context_length
+ *   3. models_dev_cache.json (built-in model database)
+ *   4. DEFAULT_CONTEXT_LENGTH (200K hardcoded fallback)
  */
 export function getModelContextLength(profile?: string): number {
   const profileDir = getProfileDir(profile)
-  const model = getDefaultModel(profileDir)
+  const config = loadConfig(profileDir)
+  if (!config) return DEFAULT_CONTEXT_LENGTH
+
+  const model = getDefaultModel(config)
   if (!model) return DEFAULT_CONTEXT_LENGTH
 
-  return lookupContextFromCache(model) || DEFAULT_CONTEXT_LENGTH
+  // 1. Global context_length override in config.yaml
+  const configCtx = getConfigContextLength(config)
+  if (configCtx && configCtx > 0) return configCtx
+
+  // 2. Custom provider context_length
+  const provider = getDefaultProvider(config)
+  const customCtx = lookupCustomProviderContextLength(config, model, provider)
+  if (customCtx && customCtx > 0) return customCtx
+
+  // 3. models_dev_cache.json
+  const cached = lookupContextFromCache(model)
+  if (cached) return cached
+
+  // 4. Fallback
+  return DEFAULT_CONTEXT_LENGTH
 }