@fastino-ai/pioneer-cli 0.1.0 → 0.2.1

package/datasets/test-remote.json

@@ -0,0 +1,41 @@
+{
+  "task_type": "ner",
+  "data": [
+    {
+      "data": {
+        "text": "Sophia spent the afternoon preparing for her biology midterm, which was scheduled for next Thursday.",
+        "entities": [
+          [
+            "biology midterm",
+            "TEST"
+          ]
+        ]
+      },
+      "text": "Sophia spent the afternoon preparing for her biology midterm, which was scheduled for next Thursday.",
+      "entities": [
+        [
+          "biology midterm",
+          "TEST"
+        ]
+      ]
+    },
+    {
+      "data": {
+        "text": "After reviewing the results from the glucose tolerance assessment, Dr. Lee recommended additional diagnostics.",
+        "entities": [
+          [
+            "glucose tolerance assessment",
+            "TEST"
+          ]
+        ]
+      },
+      "text": "After reviewing the results from the glucose tolerance assessment, Dr. Lee recommended additional diagnostics.",
+      "entities": [
+        [
+          "glucose tolerance assessment",
+          "TEST"
+        ]
+      ]
+    }
+  ]
+}

package/datasets/test-warning.json

@@ -0,0 +1,49 @@
+{
+  "task_type": "ner",
+  "data": [
+    {
+      "data": {
+        "text": "The results from the glucose tolerance assessment were consistent with previous TEST outcomes, confirming the reliability of the procedure.",
+        "entities": [
+          [
+            "TEST",
+            "TEST"
+          ]
+        ]
+      },
+      "text": "The results from the glucose tolerance assessment were consistent with previous TEST outcomes, confirming the reliability of the procedure.",
+      "entities": [
+        [
+          "TEST",
+          "TEST"
+        ]
+      ]
+    },
+    {
+      "data": {
+        "text": "After completing the midterm assessment, students will proceed to the final proficiency exam next month.",
+        "entities": [
+          [
+            "midterm assessment",
+            "TEST"
+          ],
+          [
+            "final proficiency exam",
+            "TEST"
+          ]
+        ]
+      },
+      "text": "After completing the midterm assessment, students will proceed to the final proficiency exam next month.",
+      "entities": [
+        [
+          "midterm assessment",
+          "TEST"
+        ],
+        [
+          "final proficiency exam",
+          "TEST"
+        ]
+      ]
+    }
+  ]
+}

package/ner_dataset.json

@@ -0,0 +1,111 @@
+[
+  {
+    "id": 1,
+    "text": "Apple Inc. announced that Tim Cook will visit their new facility in Cupertino, California on March 15, 2024.",
+    "entities": [
+      {"text": "Apple Inc.", "start": 0, "end": 10, "label": "ORGANIZATION"},
+      {"text": "Tim Cook", "start": 27, "end": 35, "label": "PERSON"},
+      {"text": "Cupertino", "start": 70, "end": 79, "label": "LOCATION"},
+      {"text": "California", "start": 81, "end": 91, "label": "LOCATION"},
+      {"text": "March 15, 2024", "start": 95, "end": 109, "label": "DATE"}
+    ]
+  },
+  {
+    "id": 2,
+    "text": "Dr. Sarah Johnson from Harvard University published groundbreaking research on quantum computing last Wednesday.",
+    "entities": [
+      {"text": "Sarah Johnson", "start": 4, "end": 17, "label": "PERSON"},
+      {"text": "Harvard University", "start": 23, "end": 41, "label": "ORGANIZATION"},
+      {"text": "last Wednesday", "start": 98, "end": 112, "label": "DATE"}
+    ]
+  },
+  {
+    "id": 3,
+    "text": "The European Union imposed new sanctions on Russia following the meeting in Brussels between President Biden and Emmanuel Macron.",
+    "entities": [
+      {"text": "European Union", "start": 4, "end": 18, "label": "ORGANIZATION"},
+      {"text": "Russia", "start": 44, "end": 50, "label": "LOCATION"},
+      {"text": "Brussels", "start": 76, "end": 84, "label": "LOCATION"},
+      {"text": "Biden", "start": 103, "end": 108, "label": "PERSON"},
+      {"text": "Emmanuel Macron", "start": 113, "end": 128, "label": "PERSON"}
+    ]
+  },
+  {
+    "id": 4,
+    "text": "Amazon Web Services reported $25 billion in revenue for Q4 2023, with significant growth in the Asia-Pacific region.",
+    "entities": [
+      {"text": "Amazon Web Services", "start": 0, "end": 19, "label": "ORGANIZATION"},
+      {"text": "$25 billion", "start": 29, "end": 40, "label": "MONEY"},
+      {"text": "Q4 2023", "start": 56, "end": 63, "label": "DATE"},
+      {"text": "Asia-Pacific", "start": 96, "end": 108, "label": "LOCATION"}
+    ]
+  },
+  {
+    "id": 5,
+    "text": "Professor Michael Chen will deliver a keynote speech at the International Conference on AI in Tokyo, Japan next month.",
+    "entities": [
+      {"text": "Michael Chen", "start": 10, "end": 22, "label": "PERSON"},
+      {"text": "International Conference on AI", "start": 60, "end": 90, "label": "EVENT"},
+      {"text": "Tokyo", "start": 94, "end": 99, "label": "LOCATION"},
+      {"text": "Japan", "start": 101, "end": 106, "label": "LOCATION"},
+      {"text": "next month", "start": 107, "end": 117, "label": "DATE"}
+    ]
+  },
+  {
+    "id": 6,
+    "text": "The World Health Organization confirmed 500 new cases in Mumbai, India, prompting WHO Director-General Tedros Adhanom to issue a statement.",
+    "entities": [
+      {"text": "World Health Organization", "start": 4, "end": 29, "label": "ORGANIZATION"},
+      {"text": "500", "start": 40, "end": 43, "label": "QUANTITY"},
+      {"text": "Mumbai", "start": 57, "end": 63, "label": "LOCATION"},
+      {"text": "India", "start": 65, "end": 70, "label": "LOCATION"},
+      {"text": "WHO", "start": 82, "end": 85, "label": "ORGANIZATION"},
+      {"text": "Tedros Adhanom", "start": 103, "end": 117, "label": "PERSON"}
+    ]
+  },
+  {
+    "id": 7,
+    "text": "Tesla's stock rose 15% after Elon Musk announced a new manufacturing plant in Berlin, Germany, scheduled to open in January 2025.",
+    "entities": [
+      {"text": "Tesla", "start": 0, "end": 5, "label": "ORGANIZATION"},
+      {"text": "15%", "start": 18, "end": 21, "label": "PERCENT"},
+      {"text": "Elon Musk", "start": 28, "end": 37, "label": "PERSON"},
+      {"text": "Berlin", "start": 78, "end": 84, "label": "LOCATION"},
+      {"text": "Germany", "start": 86, "end": 93, "label": "LOCATION"},
+      {"text": "January 2025", "start": 116, "end": 128, "label": "DATE"}
+    ]
+  },
+  {
+    "id": 8,
+    "text": "The Nobel Prize in Literature was awarded to British author Margaret Atwood at the ceremony in Stockholm on December 10th.",
+    "entities": [
+      {"text": "Nobel Prize in Literature", "start": 4, "end": 29, "label": "EVENT"},
+      {"text": "British", "start": 45, "end": 52, "label": "NATIONALITY"},
+      {"text": "Margaret Atwood", "start": 60, "end": 75, "label": "PERSON"},
+      {"text": "Stockholm", "start": 95, "end": 104, "label": "LOCATION"},
+      {"text": "December 10th", "start": 108, "end": 121, "label": "DATE"}
+    ]
+  },
+  {
+    "id": 9,
+    "text": "Microsoft Corporation acquired OpenAI for approximately $10 billion, as confirmed by CEO Satya Nadella in Redmond yesterday.",
+    "entities": [
+      {"text": "Microsoft Corporation", "start": 0, "end": 21, "label": "ORGANIZATION"},
+      {"text": "OpenAI", "start": 31, "end": 37, "label": "ORGANIZATION"},
+      {"text": "$10 billion", "start": 56, "end": 67, "label": "MONEY"},
+      {"text": "Satya Nadella", "start": 89, "end": 102, "label": "PERSON"},
+      {"text": "Redmond", "start": 106, "end": 113, "label": "LOCATION"},
+      {"text": "yesterday", "start": 114, "end": 123, "label": "DATE"}
+    ]
+  },
+  {
+    "id": 10,
+    "text": "Dr. Li Wei from the Beijing Institute of Technology developed a new algorithm that processes 10 terabytes of data in under 2 hours.",
+    "entities": [
+      {"text": "Li Wei", "start": 4, "end": 10, "label": "PERSON"},
+      {"text": "Beijing Institute of Technology", "start": 20, "end": 51, "label": "ORGANIZATION"},
+      {"text": "10 terabytes", "start": 93, "end": 105, "label": "QUANTITY"},
+      {"text": "2 hours", "start": 124, "end": 131, "label": "TIME"}
+    ]
+  }
+]

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@fastino-ai/pioneer-cli",
-  "version": "0.1.0",
-  "description": "Pioneer CLI - AI training platform",
+  "version": "0.2.1",
+  "description": "Pioneer CLI - AI training platform with chat agent",
   "type": "module",
   "publishConfig": {
     "access": "public"
@@ -11,13 +11,16 @@
   },
   "scripts": {
     "start": "bun run src/index.tsx",
+    "chat": "bun run src/index.tsx chat",
     "dev": "bun run --hot src/index.tsx",
     "typecheck": "tsc --noEmit"
   },
   "dependencies": {
+    "@anthropic-ai/sdk": "^0.39.0",
     "ink": "^4.4.1",
     "ink-spinner": "^5.0.0",
     "ink-text-input": "^5.0.1",
+    "openai": "^4.77.0",
     "react": "^18.3.1"
   },
   "devDependencies": {
@@ -28,4 +31,4 @@
   "engines": {
     "bun": ">=1.1.0"
   }
-}
+}

package/src/agent/Agent.ts

@@ -0,0 +1,342 @@
+/**
+ * Agent - Main orchestrator that coordinates tools and LLM calls
+ */
+
+import type {
+  Message,
+  Tool,
+  ToolCall,
+  ToolResult,
+  AgentConfig,
+  ConversationContext,
+  LLMResponse,
+  Budget,
+} from "./types.js";
+import { LLMClient } from "./LLMClient.js";
+import { ToolRegistry } from "./ToolRegistry.js";
+import { BudgetManager } from "./BudgetManager.js";
+import { FileResolver, type FileReference } from "./FileResolver.js";
+
+const DEFAULT_SYSTEM_PROMPT = `You are an intelligent AI assistant with access to powerful tools for coding, file manipulation, and system operations.
+
+Available capabilities:
+- Execute bash commands to interact with the system
+- Read, write, and edit files
+- List and search directories
+- Execute code in isolated sandboxes (Python, JavaScript, TypeScript, Bash, Ruby, Go)
+
+Guidelines:
+1. Be concise but thorough in your responses
+2. When given a task, break it down into steps and execute them
+3. Use tools to verify your work when appropriate
+4. If something fails, analyze the error and try alternative approaches
+5. Ask clarifying questions only when truly necessary
+6. Report your progress and results clearly
+
+You are operating in the user's local environment. Be careful with destructive operations.`;
+
+export interface AgentEvents {
+  onMessage?: (message: Message) => void;
+  onToolCall?: (toolCall: ToolCall) => void;
+  onToolResult?: (result: ToolResult) => void;
+  onStream?: (chunk: string) => void;
+  onBudgetWarning?: (warnings: string[]) => void;
+  onFileReference?: (refs: FileReference[]) => void;
+  onError?: (error: Error) => void;
+}
+
+export class Agent {
+  private llm: LLMClient;
+  private tools: ToolRegistry;
+  private budget: BudgetManager;
+  private config: AgentConfig;
+  private context: ConversationContext;
+  private events: AgentEvents;
+  private fileResolver: FileResolver;
+  private abortController: AbortController | null = null;
+
+  constructor(config: AgentConfig, events: AgentEvents = {}) {
+    this.config = config;
+    this.events = events;
+
+    this.llm = new LLMClient({
+      provider: config.provider,
+      model: config.model,
+      apiKey: config.apiKey,
+      baseUrl: config.baseUrl,
+    });
+
+    this.tools = new ToolRegistry();
+    this.budget = new BudgetManager(
+      config.budget,
+      this.llm.getCostPerMillion()
+    );
+    this.fileResolver = new FileResolver(process.cwd());
+
+    this.context = {
+      messages: [],
+      budgetUsage: this.budget.getUsage(),
+      startTime: new Date(),
+      workingDirectory: process.cwd(),
+    };
+  }
+
+  registerTool(tool: Tool): void {
+    this.tools.register(tool);
+  }
+
+  registerTools(tools: Tool[]): void {
+    for (const tool of tools) {
+      this.tools.register(tool);
+    }
+  }
+
+  getTools(): Tool[] {
+    return this.tools.getAll();
+  }
+
+  async chat(userMessage: string, stream = true): Promise<string> {
+    // Create new abort controller for this chat session
+    this.abortController = new AbortController();
+
+    // Check budget before starting
+    const budgetCheck = this.budget.checkBudget();
+    if (!budgetCheck.withinBudget) {
+      const error = `Budget exhausted: ${budgetCheck.warnings.join(", ")}`;
+      this.events.onError?.(new Error(error));
+      return error;
+    }
+
+    if (budgetCheck.warnings.length > 0) {
+      this.events.onBudgetWarning?.(budgetCheck.warnings);
+    }
+
+    // Resolve @file references in the message
+    const resolved = this.fileResolver.resolve(userMessage);
+
+    // Notify about file references
+    if (resolved.references.length > 0) {
+      this.events.onFileReference?.(resolved.references);
+    }
+
+    // Build the actual message content with file context
+    let messageContent = userMessage;
+    if (resolved.contextBlock) {
+      messageContent = resolved.contextBlock + "\n" + userMessage;
+    }
+
+    // Add user message to context
+    const userMsg: Message = {
+      role: "user",
+      content: messageContent,
+      timestamp: new Date(),
+    };
+    this.context.messages.push(userMsg);
+    this.events.onMessage?.(userMsg);
+
+    // Run the agent loop
+    try {
+      return await this.runAgentLoop(stream);
+    } finally {
+      this.abortController = null;
+    }
+  }
+
+  // Stop the current generation/tool execution
+  stop(): void {
+    if (this.abortController) {
+      this.abortController.abort();
+    }
+  }
+
+  // Check if agent is currently processing
+  isProcessing(): boolean {
+    return this.abortController !== null;
+  }
+
+  private async runAgentLoop(stream: boolean): Promise<string> {
+    const maxToolCalls = this.config.maxToolCalls || 25;
+    let toolCallCount = 0;
+    let finalResponse = "";
+
+    while (toolCallCount < maxToolCalls) {
+      // Check if aborted
+      if (this.abortController?.signal.aborted) {
+        finalResponse += "\n\n[Stopped by user]";
+        break;
+      }
+
+      // Check budget
+      const budgetCheck = this.budget.checkBudget();
+      if (!budgetCheck.withinBudget) {
+        finalResponse += `\n\n[Budget exhausted: ${budgetCheck.warnings.join(", ")}]`;
+        break;
+      }
+
+      if (budgetCheck.warnings.length > 0) {
+        this.events.onBudgetWarning?.(budgetCheck.warnings);
+      }
+
+      // Get LLM response
+      const systemPrompt = this.config.systemPrompt || DEFAULT_SYSTEM_PROMPT;
+
+      let response: LLMResponse;
+      try {
+        response = await this.llm.chat(
+          this.context.messages,
+          this.tools.getAll(),
+          systemPrompt,
+          stream ? this.events.onStream : undefined,
+          this.abortController?.signal
+        );
+      } catch (error) {
+        const err = error instanceof Error ? error : new Error(String(error));
+        // Don't report abort as an error
+        if (err.message === "Aborted") {
+          return finalResponse + "\n\n[Stopped by user]";
+        }
+        this.events.onError?.(err);
+        return `Error communicating with LLM: ${err.message}`;
+      }
+
+      // Record token usage
+      this.budget.recordUsage(response.usage);
+      this.budget.recordIteration();
+
+      // Add assistant message to context
+      const assistantMsg: Message = {
+        role: "assistant",
+        content: response.content,
+        toolCalls: response.toolCalls,
+        timestamp: new Date(),
+      };
+      this.context.messages.push(assistantMsg);
+      this.events.onMessage?.(assistantMsg);
+
+      // If no tool calls, we're done
+      if (!response.toolCalls || response.toolCalls.length === 0) {
+        finalResponse = response.content;
+        break;
+      }
+
+      // Execute tool calls
+      for (const toolCall of response.toolCalls) {
+        // Check if aborted before each tool call
+        if (this.abortController?.signal.aborted) {
+          return finalResponse + "\n\n[Stopped by user]";
+        }
+
+        this.events.onToolCall?.(toolCall);
+        toolCallCount++;
+
+        const result = await this.executeToolCall(toolCall);
+        this.events.onToolResult?.(result);
+
+        // Add tool result to context
+        const toolMsg: Message = {
+          role: "tool",
+          content: result.output || result.error || "(no output)",
+          toolCallId: toolCall.id,
+          timestamp: new Date(),
+        };
+        this.context.messages.push(toolMsg);
+      }
+
+      // If we hit max tool calls, break
+      if (toolCallCount >= maxToolCalls) {
+        finalResponse += `\n\n[Reached maximum tool call limit: ${maxToolCalls}]`;
+        break;
+      }
+    }
+
+    return finalResponse;
+  }
+
+  private async executeToolCall(toolCall: ToolCall): Promise<ToolResult> {
+    const result = await this.tools.execute(toolCall.name, toolCall.arguments);
+    result.toolCallId = toolCall.id;
+    return result;
+  }
+
+  // Get current conversation
+  getMessages(): Message[] {
+    return [...this.context.messages];
+  }
+
+  // Clear conversation history
+  clearHistory(): void {
+    this.context.messages = [];
+    this.budget.reset();
+    this.context.startTime = new Date();
+  }
+
+  // Get budget status
+  getBudgetStatus(): {
+    usage: ReturnType<BudgetManager["getUsage"]>;
+    check: ReturnType<BudgetManager["checkBudget"]>;
+    summary: string;
+  } {
+    return {
+      usage: this.budget.getUsage(),
+      check: this.budget.checkBudget(),
+      summary: this.budget.formatUsageSummary(),
+    };
+  }
+
+  // Update budget
+  updateBudget(budget: Partial<Budget>): void {
+    this.budget.updateBudget(budget);
+  }
+
+  // Update max tool calls
+  setMaxToolCalls(limit: number): void {
+    this.config.maxToolCalls = limit;
+  }
+
+  // Get max tool calls
+  getMaxToolCalls(): number {
+    return this.config.maxToolCalls || 50;
+  }
+
+  // Get current model info
+  getModelInfo(): { provider: string; model: string } {
+    return {
+      provider: this.config.provider,
+      model: this.llm.getModel(),
+    };
+  }
+
+  // Set model
+  setModel(model: string): void {
+    this.config.model = model;
+    this.llm.setModel(model);
+    // Update cost estimation for new model
+    this.budget.setCostPerMillion(this.llm.getCostPerMillion());
+  }
+
+  // Set working directory
+  setWorkingDirectory(dir: string): void {
+    this.context.workingDirectory = dir;
+    this.fileResolver.setBasePath(dir);
+  }
+
+  // Get working directory
+  getWorkingDirectory(): string {
+    return this.context.workingDirectory;
+  }
+
+  // Add a system message
+  addSystemMessage(content: string): void {
+    this.context.messages.push({
+      role: "system",
+      content,
+      timestamp: new Date(),
+    });
+  }
+
+  // Get file suggestions for autocomplete
+  getFileSuggestions(partial: string): string[] {
+    return this.fileResolver.getSuggestions(partial);
+  }
+}
+