Files
PhoneClaw/LLM/MLX/Core/MLXTokenizersLoader.swift
KellyvvandClaude Opus 4.6 779be78843 refactor: split MLX service, introduce skill type & runtime budgets
Major restructuring of the agent stack to support content-type skills,
data-driven runtime budgets, and the new translate skill.

LLM/MLX/
- Split MLXLocalLLMService.swift (1500 → 700 lines) into Core/, Gemma4/,
  Installation/ subdirectories
- New BundledModel: LinearBudgetFormula replaces tier tables, model
  parameters fully data-driven via ModelRuntimeProfile
- New RuntimeBudgets: pure functions for text/thinking/multimodal budgets,
  zero hardcoded model IDs
- Drop adjustedTextOutputTokens prepared-subtraction (chunked prefill made
  it obsolete and triggered false-positive truncation on long prompts)

Skills/
- Add SkillType (.device / .content) metadata, parsed from SKILL.md
  frontmatter
- New translate skill (type: content, no tools) with anaphora handling
- Trim translate (4563→1804), device (2357→1680), calendar (2169→1790)
- contacts SKILL.md adds multi-turn disambiguation + cancel guidance,
  semantics owned by body not framework
- SkillRegistry validates allowed-tools against ToolRegistry at load time

Agent/
- ToolCallParser becomes registry-aware: drops tool_call names not in
  ToolRegistry or protocol set (load_skill / list_skills); ToolRegistry
  is now the single white list, validation moves to parser boundary
- Router sticky routing uses semantic user-turn boundary instead of
  fixed suffix(4); correctly handles long agent loops
- AgentEngine system prompt template differentiates device vs content
  skills with explicit invocation rules
- SYSPROMPT.md auto-migration: old single-placeholder template backed up
  to .bak when new dual-placeholder is detected
- PromptBuilder.buildLoadedSkillPrompt receives availableTools list,
  explicitly tells model which tools (or none) are callable

Tools/
- ToolRegistry split into Helpers/Stores/AppPermissions/Handlers/* for
  separation of concerns
- Mark Tools/Handlers/* with explicit allowed argument signatures

UI/
- Add long-press copy contextMenu on user and assistant bubbles
- Enable .textSelection on user bubble

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-08 09:22:20 +08:00

56 lines
1.7 KiB
Swift

import Foundation
import MLXLMCommon
import Tokenizers
/// Bridges `swift-tokenizers` to `MLXLMCommon.Tokenizer`.
struct MLXTokenizerBridge: MLXLMCommon.Tokenizer {
private let upstream: any Tokenizers.Tokenizer
init(_ upstream: any Tokenizers.Tokenizer) {
self.upstream = upstream
}
func encode(text: String, addSpecialTokens: Bool) -> [Int] {
upstream.encode(text: text, addSpecialTokens: addSpecialTokens)
}
func decode(tokenIds: [Int], skipSpecialTokens: Bool) -> String {
upstream.decode(tokenIds: tokenIds, skipSpecialTokens: skipSpecialTokens)
}
func convertTokenToId(_ token: String) -> Int? {
upstream.convertTokenToId(token)
}
func convertIdToToken(_ id: Int) -> String? {
upstream.convertIdToToken(id)
}
var bosToken: String? { upstream.bosToken }
var eosToken: String? { upstream.eosToken }
var unknownToken: String? { upstream.unknownToken }
func applyChatTemplate(
messages: [[String: any Sendable]],
tools: [[String: any Sendable]]?,
additionalContext: [String: any Sendable]?
) throws -> [Int] {
do {
return try upstream.applyChatTemplate(
messages: messages,
tools: tools,
additionalContext: additionalContext
)
} catch let error as Tokenizers.TokenizerError where error == .missingChatTemplate {
throw MLXLMCommon.TokenizerError.missingChatTemplate
}
}
}
struct MLXTokenizersLoader: TokenizerLoader {
func load(from directory: URL) async throws -> any MLXLMCommon.Tokenizer {
let upstream = try await AutoTokenizer.from(directory: directory)
return MLXTokenizerBridge(upstream)
}
}