activo 0.4.4 → 0.5.0

package/src/core/llm/ollama.ts

@@ -1,351 +0,0 @@
-import { OllamaConfig } from "../config.js";
-import type { Tool, ToolCall } from "../tools/types.js";
-
-export interface ChatMessage {
-  role: "system" | "user" | "assistant" | "tool";
-  content: string;
-  toolCalls?: ToolCall[];
-  toolCallId?: string;
-}
-
-export interface StreamEvent {
-  type: "content" | "tool_call" | "done" | "error";
-  content?: string;
-  toolCall?: ToolCall;
-  error?: string;
-}
-
-interface OllamaChatResponse {
-  model: string;
-  message: {
-    role: string;
-    content: string;
-    tool_calls?: Array<{
-      function: {
-        name: string;
-        arguments: Record<string, unknown>;
-      };
-    }>;
-  };
-  done: boolean;
-}
-
-// Estimate token count for a string (rough: ~3 chars per token for mixed Korean/English)
-function estimateTokens(text: string): number {
-  if (!text) return 0;
-  return Math.ceil(text.length / 3);
-}
-
-// Estimate tokens for tool definitions sent to Ollama
-function estimateToolTokens(tools: Tool[]): number {
-  let total = 0;
-  for (const tool of tools) {
-    // name + description + JSON schema overhead
-    total += estimateTokens(tool.name) + estimateTokens(tool.description) + 20;
-    if (tool.parameters?.properties) {
-      for (const [key, prop] of Object.entries(tool.parameters.properties)) {
-        total += estimateTokens(key) + estimateTokens((prop as { description?: string }).description || "") + 10;
-      }
-    }
-  }
-  return total;
-}
-
-// Prune messages to fit within context window
-function pruneMessages(
-  messages: Array<{ role: string; content: string }>,
-  maxContextTokens: number,
-  toolTokens: number
-): Array<{ role: string; content: string }> {
-  if (messages.length <= 2) return messages;
-
-  const responseReserve = 1000; // reserve tokens for model response
-  const safetyBuffer = 200;
-  const budget = maxContextTokens - toolTokens - responseReserve - safetyBuffer;
-
-  // Always preserve: first message (system) and last message (user)
-  const systemMsg = messages[0];
-  const lastMsg = messages[messages.length - 1];
-  const history = messages.slice(1, -1);
-
-  let usedTokens = estimateTokens(systemMsg.content) + estimateTokens(lastMsg.content);
-
-  // If even system + last message exceeds budget, truncate system prompt
-  if (usedTokens > budget) {
-    const maxSystemChars = Math.max(200, (budget - estimateTokens(lastMsg.content)) * 3);
-    systemMsg.content = systemMsg.content.slice(0, maxSystemChars);
-    return [systemMsg, lastMsg];
-  }
-
-  // Add messages from newest to oldest
-  const kept: Array<{ role: string; content: string }> = [];
-  for (let i = history.length - 1; i >= 0; i--) {
-    const msgTokens = estimateTokens(history[i].content) + 4; // 4 tokens overhead per message
-    if (usedTokens + msgTokens > budget) break;
-    kept.unshift(history[i]);
-    usedTokens += msgTokens;
-  }
-
-  return [systemMsg, ...kept, lastMsg];
-}
-
-export class OllamaClient {
-  private baseUrl: string;
-  private model: string;
-  private contextLength: number;
-  private keepAlive: number;
-
-  constructor(config: OllamaConfig) {
-    this.baseUrl = config.baseUrl;
-    this.model = config.model;
-    this.contextLength = config.contextLength;
-    this.keepAlive = config.keepAlive;
-  }
-
-  async isConnected(): Promise<boolean> {
-    try {
-      const response = await fetch(`${this.baseUrl}/api/tags`);
-      return response.ok;
-    } catch {
-      return false;
-    }
-  }
-
-  async listModels(): Promise<string[]> {
-    const response = await fetch(`${this.baseUrl}/api/tags`);
-    if (!response.ok) throw new Error("Failed to list models");
-    const data = (await response.json()) as { models: Array<{ name: string }> };
-    return data.models.map((m) => m.name);
-  }
-
-  async chat(
-    messages: ChatMessage[],
-    tools?: Tool[]
-  ): Promise<ChatMessage> {
-    let ollamaMessages = this.convertMessages(messages);
-
-    // Prune messages to fit within context window
-    const toolTokens = tools?.length ? estimateToolTokens(tools) : 0;
-    ollamaMessages = pruneMessages(ollamaMessages, this.contextLength, toolTokens);
-
-    const body: Record<string, unknown> = {
-      model: this.model,
-      messages: ollamaMessages,
-      stream: false,
-      options: {
-        num_ctx: this.contextLength,
-      },
-      keep_alive: this.keepAlive,
-    };
-
-    // Always add tools if provided (needed for multi-turn tool use)
-    if (tools?.length) {
-      body.tools = tools.map((tool) => ({
-        type: "function",
-        function: {
-          name: tool.name,
-          description: tool.description,
-          parameters: tool.parameters,
-        },
-      }));
-    }
-
-    const response = await fetch(`${this.baseUrl}/api/chat`, {
-      method: "POST",
-      headers: { "Content-Type": "application/json" },
-      body: JSON.stringify(body),
-    });
-
-    if (!response.ok) {
-      const error = await response.text();
-      throw new Error(`Ollama error: ${error}`);
-    }
-
-    const data = (await response.json()) as OllamaChatResponse;
-    return this.parseResponse(data);
-  }
-
-  async *streamChat(
-    messages: ChatMessage[],
-    tools?: Tool[],
-    abortSignal?: AbortSignal
-  ): AsyncGenerator<StreamEvent> {
-    let ollamaMessages = this.convertMessages(messages);
-
-    // Prune messages to fit within context window
-    const toolTokens = tools?.length ? estimateToolTokens(tools) : 0;
-    ollamaMessages = pruneMessages(ollamaMessages, this.contextLength, toolTokens);
-
-    // Use non-streaming mode when tools are provided to avoid hallucination
-    // (LLM often outputs fake results before tool calls in streaming mode)
-    const useStreaming = !tools?.length;
-
-    const body: Record<string, unknown> = {
-      model: this.model,
-      messages: ollamaMessages,
-      stream: useStreaming,
-      options: {
-        num_ctx: this.contextLength,
-      },
-      keep_alive: this.keepAlive,
-    };
-
-    // Always add tools if provided (needed for multi-turn tool use)
-    if (tools?.length) {
-      body.tools = tools.map((tool) => ({
-        type: "function",
-        function: {
-          name: tool.name,
-          description: tool.description,
-          parameters: tool.parameters,
-        },
-      }));
-    }
-
-    const response = await fetch(`${this.baseUrl}/api/chat`, {
-      method: "POST",
-      headers: { "Content-Type": "application/json" },
-      body: JSON.stringify(body),
-      signal: abortSignal,
-    });
-
-    if (!response.ok) {
-      const error = await response.text();
-      yield { type: "error", error: `Ollama error: ${error}` };
-      return;
-    }
-
-    // Non-streaming mode for tool calls (avoids hallucination where LLM outputs text before calling tools)
-    if (!useStreaming) {
-      const data = (await response.json()) as OllamaChatResponse;
-
-      // Handle tool calls first (ignore content when tools are called)
-      if (data.message?.tool_calls?.length) {
-        for (const tc of data.message.tool_calls) {
-          const toolCall: ToolCall = {
-            id: `tc_${Date.now()}_${Math.random().toString(36).slice(2)}`,
-            name: tc.function.name,
-            arguments: tc.function.arguments,
-          };
-          yield { type: "tool_call", toolCall };
-        }
-      } else if (data.message?.content) {
-        // Only yield content if no tool calls
-        yield { type: "content", content: data.message.content };
-      }
-
-      yield { type: "done" };
-      return;
-    }
-
-    // Streaming mode (no tools)
-    const reader = response.body?.getReader();
-    if (!reader) {
-      yield { type: "error", error: "No response body" };
-      return;
-    }
-
-    const decoder = new TextDecoder();
-    let buffer = "";
-
-    while (true) {
-      const { done, value } = await reader.read();
-      if (done) break;
-
-      buffer += decoder.decode(value, { stream: true });
-      const lines = buffer.split("\n");
-      buffer = lines.pop() || "";
-
-      for (const line of lines) {
-        if (!line.trim()) continue;
-
-        try {
-          const data = JSON.parse(line) as OllamaChatResponse;
-
-          if (data.message?.content) {
-            yield { type: "content", content: data.message.content };
-          }
-
-          if (data.done) {
-            yield { type: "done" };
-          }
-        } catch {
-          // Skip invalid JSON
-        }
-      }
-    }
-  }
-
-  private convertMessages(messages: ChatMessage[]): Array<{
-    role: string;
-    content: string;
-  }> {
-    return messages.map((msg) => {
-      if (msg.role === "tool") {
-        return {
-          role: "tool",
-          content: msg.content,
-        };
-      }
-      return {
-        role: msg.role,
-        content: msg.content,
-      };
-    });
-  }
-
-  private parseResponse(data: OllamaChatResponse): ChatMessage {
-    const result: ChatMessage = {
-      role: "assistant",
-      content: data.message.content || "",
-    };
-
-    if (data.message.tool_calls?.length) {
-      result.toolCalls = data.message.tool_calls.map((tc, idx) => ({
-        id: `tc_${Date.now()}_${idx}`,
-        name: tc.function.name,
-        arguments: tc.function.arguments,
-      }));
-    }
-
-    return result;
-  }
-
-  getModel(): string {
-    return this.model;
-  }
-
-  setModel(model: string): void {
-    this.model = model;
-  }
-
-  // Generate embeddings for text
-  async embed(text: string, model?: string): Promise<number[]> {
-    const response = await fetch(`${this.baseUrl}/api/embeddings`, {
-      method: "POST",
-      headers: { "Content-Type": "application/json" },
-      body: JSON.stringify({
-        model: model || "nomic-embed-text",
-        prompt: text,
-      }),
-    });
-
-    if (!response.ok) {
-      const error = await response.text();
-      throw new Error(`Ollama embedding error: ${error}`);
-    }
-
-    const data = (await response.json()) as { embedding: number[] };
-    return data.embedding;
-  }
-
-  // Generate embeddings for multiple texts (batch)
-  async embedBatch(texts: string[], model?: string): Promise<number[][]> {
-    const embeddings: number[][] = [];
-    for (const text of texts) {
-      const embedding = await this.embed(text, model);
-      embeddings.push(embedding);
-    }
-    return embeddings;
-  }
-}

package/src/core/mcp/client.ts

@@ -1,143 +0,0 @@
-import { Client } from "@modelcontextprotocol/sdk/client/index.js";
-import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
-import { MCPServerConfig } from "../config.js";
-import { Tool, ToolResult } from "../tools/types.js";
-
-export interface MCPConnection {
-  id: string;
-  client: Client;
-  transport: StdioClientTransport;
-  tools: Tool[];
-}
-
-export class MCPManager {
-  private connections: Map<string, MCPConnection> = new Map();
-
-  async connect(id: string, config: MCPServerConfig): Promise<MCPConnection> {
-    // Check if already connected
-    if (this.connections.has(id)) {
-      return this.connections.get(id)!;
-    }
-
-    const transport = new StdioClientTransport({
-      command: config.command,
-      args: config.args,
-      env: config.env,
-    });
-
-    const client = new Client({
-      name: "activo",
-      version: "0.2.0",
-    });
-
-    await client.connect(transport);
-
-    // Get available tools
-    const toolsResult = await client.listTools();
-    const tools: Tool[] = toolsResult.tools.map((t) => ({
-      name: `mcp_${id}_${t.name}`,
-      description: t.description || `MCP tool: ${t.name}`,
-      parameters: {
-        type: "object" as const,
-        properties: (t.inputSchema as any)?.properties || {},
-        required: (t.inputSchema as any)?.required,
-      },
-      handler: async (args: Record<string, unknown>): Promise<ToolResult> => {
-        return this.callTool(id, t.name, args);
-      },
-    }));
-
-    const connection: MCPConnection = {
-      id,
-      client,
-      transport,
-      tools,
-    };
-
-    this.connections.set(id, connection);
-    return connection;
-  }
-
-  async disconnect(id: string): Promise<void> {
-    const connection = this.connections.get(id);
-    if (connection) {
-      await connection.client.close();
-      this.connections.delete(id);
-    }
-  }
-
-  async disconnectAll(): Promise<void> {
-    for (const id of this.connections.keys()) {
-      await this.disconnect(id);
-    }
-  }
-
-  async callTool(connectionId: string, toolName: string, args: Record<string, unknown>): Promise<ToolResult> {
-    const connection = this.connections.get(connectionId);
-    if (!connection) {
-      return {
-        success: false,
-        content: "",
-        error: `MCP connection not found: ${connectionId}`,
-      };
-    }
-
-    try {
-      const result = await connection.client.callTool({
-        name: toolName,
-        arguments: args,
-      });
-
-      if (result.isError) {
-        return {
-          success: false,
-          content: "",
-          error: JSON.stringify(result.content),
-        };
-      }
-
-      const contentArray = result.content as Array<{ type: string; text?: string }>;
-      const content = contentArray
-        .filter((c) => c.type === "text" && c.text)
-        .map((c) => c.text!)
-        .join("\n");
-
-      return {
-        success: true,
-        content,
-      };
-    } catch (error) {
-      return {
-        success: false,
-        content: "",
-        error: String(error),
-      };
-    }
-  }
-
-  getAllTools(): Tool[] {
-    const tools: Tool[] = [];
-    for (const connection of this.connections.values()) {
-      tools.push(...connection.tools);
-    }
-    return tools;
-  }
-
-  getConnection(id: string): MCPConnection | undefined {
-    return this.connections.get(id);
-  }
-
-  isConnected(id: string): boolean {
-    return this.connections.has(id);
-  }
-}
-
-// Singleton instance
-let mcpManager: MCPManager | null = null;
-
-export function getMCPManager(): MCPManager {
-  if (!mcpManager) {
-    mcpManager = new MCPManager();
-  }
-  return mcpManager;
-}