jinzd-ai-cli 0.2.9 → 0.2.11

package/README.md

@@ -9,6 +9,7 @@
 [![Node.js](https://img.shields.io/badge/node-%3E%3D20-brightgreen)](https://nodejs.org)
 [![Tests](https://img.shields.io/badge/tests-138%20passing-brightgreen)]()
 [![GitHub Release](https://img.shields.io/github/v/release/jinzhengdong/ai-cli)](https://github.com/jinzhengdong/ai-cli/releases)
+[![CI](https://github.com/jinzhengdong/ai-cli/actions/workflows/ci.yml/badge.svg)](https://github.com/jinzhengdong/ai-cli/actions/workflows/ci.yml)
 
 **ai-cli** is a powerful AI assistant that connects to 7 providers and executes tasks autonomously through agentic tool calling. Use it as a terminal REPL, a browser-based Web UI, or a standalone Electron desktop app.
 
@@ -26,13 +27,15 @@
 - **Streaming Tool Use** — Real-time streaming of AI reasoning and tool calls as they happen
 - **Sub-Agents** — Delegate complex subtasks to isolated child agents with independent tool loops
 - **Extended Thinking** — Claude deep reasoning mode with `/think` toggle
-- **Plan Mode** — Read-only planning phase (`/plan`) where AI analyzes before executing
+- **Plan Mode** — Read-only planning phase (`/plan`) where AI analyzes before executing, with loop detection
+- **Auto-Pause** — Automatically pauses every 10 rounds for user review and redirection
 - **MCP Protocol** — Connect external MCP servers for dynamic tool discovery
 - **Multi-User Auth** — Web UI supports multiple users with password authentication
 - **PWA Support** — Install Web UI as a desktop/mobile app, accessible over LAN
 - **Hierarchical Context** — 3-layer context files (global / project / subdirectory) auto-injected
 - **Headless Mode** — `ai-cli -p "prompt"` for CI/CD pipelines and scripting
 - **35+ REPL Commands** — Session management, checkpointing, code review, scaffolding, and more
+- **GitHub Actions CI/CD** — Automated testing on Node 20/22 + npm publish on release tags
 - **Cross-Platform** — Windows, macOS, Linux
 
 ## Installation
@@ -125,7 +128,7 @@ AI autonomously invokes these 16 tools during conversations:
 | `bash` | varies | Execute shell commands (PowerShell on Windows, $SHELL on Unix) |
 | `read_file` | safe | Read file contents (10 MB limit, image support) |
 | `write_file` | write | Create/overwrite files (diff preview + confirmation) |
-| `edit_file` | write | Precise string replacement with fuzzy matching hints |
+| `edit_file` | write | Precise string replacement with fuzzy matching hints + `replaceAll` mode |
 | `list_dir` | safe | List directory contents |
 | `grep_files` | safe | Regex search across files |
 | `glob_files` | safe | Match files by glob pattern |
@@ -167,6 +170,8 @@ AI autonomously invokes these 16 tools during conversations:
 | `/config` | Open configuration wizard |
 | `/help` | Show all available commands |
 
+**Multi-line input**: Use `\` at end of line for continuation, or paste multi-line content directly (auto-detected and merged).
+
 See the [full command reference](USAGE.md) for all 35+ commands.
 
 ## CLI Parameters

package/README.zh-CN.md

@@ -9,6 +9,7 @@
 [![Node.js](https://img.shields.io/badge/node-%3E%3D20-brightgreen)](https://nodejs.org)
 [![Tests](https://img.shields.io/badge/tests-138%20passing-brightgreen)]()
 [![GitHub Release](https://img.shields.io/github/v/release/jinzhengdong/ai-cli)](https://github.com/jinzhengdong/ai-cli/releases)
+[![CI](https://github.com/jinzhengdong/ai-cli/actions/workflows/ci.yml/badge.svg)](https://github.com/jinzhengdong/ai-cli/actions/workflows/ci.yml)
 
 ## 特性亮点
 
@@ -18,13 +19,15 @@
 - **流式工具调用** — 实时流式展示 AI 推理过程和工具调用
 - **子代理系统** — 将复杂子任务委派给独立子代理执行
 - **深度推理** — Claude Extended Thinking，`/think` 一键切换
-- **规划模式** — `/plan` 进入只读规划，AI 先分析后执行
+- **规划模式** — `/plan` 进入只读规划，AI 先分析后执行，内置死循环检测
+- **自动暂停** — 每 10 轮自动暂停，用户可审查进度或重定向 AI
 - **MCP 协议** — 接入外部 MCP 服务器，动态发现工具
 - **多用户认证** — Web UI 支持多用户密码登录
 - **PWA 支持** — Web UI 可安装为桌面/移动应用，支持局域网访问
 - **三层级上下文** — 全局 / 项目 / 子目录上下文文件自动注入
 - **无头模式** — `aicli -p "提示词"` 用于 CI/CD 管道和脚本
 - **35+ REPL 命令** — 会话管理、检查点、代码审查、脚手架等
+- **GitHub Actions CI/CD** — Node 20/22 自动测试 + Release tag 自动发布 npm
 - **跨平台** — Windows、macOS、Linux
 
 ## 安装
@@ -117,7 +120,7 @@ AI 在对话中可自主调用 16 个工具：
 | `bash` | 动态判断 | 执行 shell 命令（Windows: PowerShell，Unix: $SHELL） |
 | `read_file` | 安全 | 读取文件内容（10 MB 限制，支持图片） |
 | `write_file` | 写入 | 创建/覆盖文件（diff 预览 + 确认） |
-| `edit_file` | 写入 | 精确字符串替换，模糊匹配提示 |
+| `edit_file` | 写入 | 精确字符串替换，模糊匹配提示 + `replaceAll` 全局替换模式 |
 | `list_dir` | 安全 | 列出目录内容 |
 | `grep_files` | 安全 | 正则搜索文件内容 |
 | `glob_files` | 安全 | 按 glob 模式匹配文件 |
@@ -159,6 +162,8 @@ AI 在对话中可自主调用 16 个工具：
 | `/config` | 打开配置向导 |
 | `/help` | 显示所有命令 |
 
+**多行输入**：行末加 `\` 续行，或直接粘贴多行内容（自动检测合并）。
+
 完整命令参考见 [使用说明](USAGE.md)。
 
 ## CLI 参数

package/dist/{chunk-UR5QRBEG.js → chunk-HLR3T3PP.js}

@@ -16,7 +16,7 @@ import {
   SUBAGENT_MAX_ROUNDS_LIMIT,
   VERSION,
   runTestsTool
-} from "./chunk-J6SXTNJO.js";
+} from "./chunk-NT7GYZEL.js";
 
 // src/config/config-manager.ts
 import { readFileSync, writeFileSync, existsSync, mkdirSync } from "fs";
@@ -1482,7 +1482,8 @@ var DeepSeekProvider = class extends OpenAICompatibleProvider {
    */
   async chatWithTools(request, tools) {
     const result = await super.chatWithTools(request, tools);
-    if ("content" in result && result.content && detectsCodeBlockPseudoCall(result.content)) {
+    const hasBashTool = tools.some((t) => t.name === "bash");
+    if (hasBashTool && "content" in result && result.content && detectsCodeBlockPseudoCall(result.content)) {
       process.stderr.write(
         `[deepseek] \u26A0 Detected code block pseudo-tool-call (DeepSeek wrote code in text instead of calling a tool). Forcing retry...
 `

package/dist/{chunk-J6SXTNJO.js → chunk-NT7GYZEL.js}

@@ -8,7 +8,7 @@ import { platform } from "os";
 import chalk from "chalk";
 
 // src/core/constants.ts
-var VERSION = "0.2.9";
+var VERSION = "0.2.11";
 var APP_NAME = "ai-cli";
 var CONFIG_DIR_NAME = ".aicli";
 var CONFIG_FILE_NAME = "config.json";
@@ -54,6 +54,13 @@ You are currently in read-only planning (Plan) mode.
    - Execution order and dependencies
    - Potential risks and considerations
 
+**CRITICAL RULES**:
+- Do NOT attempt to call bash, write_file, edit_file, or any disabled tool \u2014 they will fail silently.
+- Do NOT write shell commands, SQL queries, or code in your text response as a substitute for tool calls \u2014 the user's system will misinterpret this as a pseudo-tool-call error.
+- If the user asks you to run commands, test connections, or modify files, respond with: "This requires execution tools. Please type \`/plan execute\` to switch to execute mode, then I can perform these operations."
+- Do NOT call write_todos repeatedly with the same content \u2014 call it once, then give a text response.
+- Focus your analysis on reading files and producing actionable plans.
+
 Once planning is complete, clearly inform the user: type \`/plan execute\` to begin executing the plan, or \`/plan exit\` to discard it.`;
 var SUBAGENT_DEFAULT_MAX_ROUNDS = 10;
 var SUBAGENT_MAX_ROUNDS_LIMIT = 15;

package/dist/index.js

@@ -35,7 +35,7 @@ import {
   theme,
   truncateOutput,
   undoStack
-} from "./chunk-UR5QRBEG.js";
+} from "./chunk-HLR3T3PP.js";
 import {
   AGENTIC_BEHAVIOR_GUIDELINE,
   AUTHOR,
@@ -55,7 +55,7 @@ import {
   REPO_URL,
   SKILLS_DIR_NAME,
   VERSION
-} from "./chunk-J6SXTNJO.js";
+} from "./chunk-NT7GYZEL.js";
 
 // src/index.ts
 import { program } from "commander";
@@ -1904,7 +1904,7 @@ ${hint}` : "")
       description: "Run project tests and show structured report",
       usage: "/test [command|filter]",
       async execute(args, _ctx) {
-        const { executeTests } = await import("./run-tests-ITCKU6DW.js");
+        const { executeTests } = await import("./run-tests-TGHLGADJ.js");
         const argStr = args.join(" ").trim();
         let testArgs = {};
         if (argStr) {
@@ -3311,7 +3311,8 @@ ${content}
 }
 var MAX_TOOL_ROUNDS = 25;
 var FREE_ROUND_TOOLS = /* @__PURE__ */ new Set(["write_todos"]);
-var MAX_CONSECUTIVE_FREE_ROUNDS = 5;
+var MAX_CONSECUTIVE_FREE_ROUNDS = 3;
+var MAX_REPEATED_TOOL_CALLS = 2;
 var AUTO_PAUSE_INTERVAL = 10;
 function fmtTokens(n) {
   if (n >= 1e6) return `${Math.round(n / 1e5) / 10}M`;
@@ -4818,6 +4819,8 @@ You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds for this task. Plan ef
     const roundUsage = { inputTokens: 0, outputTokens: 0 };
     const supportsStreamingTools = useStreaming && typeof provider.chatWithToolsStream === "function";
     let consecutiveFreeRounds = 0;
+    let lastToolCallSignature = "";
+    let repeatedToolCallCount = 0;
     this.setupInterjectionListener();
     process.stdout.write(theme.dim("  (Type a message + Enter to redirect AI at any time)\n"));
     try {
@@ -4901,7 +4904,8 @@ You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds for this task. Plan ef
         if ("content" in result) {
           const hasWriteTools = toolDefs.some((t) => t.name === "write_file" || t.name === "edit_file");
           const alreadyWrote = hadPreviousWriteToolCalls(extraMessages);
-          if (hasWriteTools && !alreadyWrote && result.content && detectsHallucinatedFileOp(result.content) && round < MAX_TOOL_ROUNDS - 1) {
+          if (!this.planMode && // Plan Mode 下跳过虚假声明检测
+          hasWriteTools && !alreadyWrote && result.content && detectsHallucinatedFileOp(result.content) && round < MAX_TOOL_ROUNDS - 1) {
             const providerName = this.currentProvider;
             process.stderr.write(
               `[${providerName}] \u26A0 Hallucinated completion detected (AI claimed file was written but no tool was called), forcing retry...
@@ -5043,6 +5047,28 @@ You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds for this task. Plan ef
         } else {
           consecutiveFreeRounds = 0;
         }
+        const currentSignature = result.toolCalls.map((tc) => `${tc.name}:${JSON.stringify(tc.arguments)}`).join("|");
+        if (currentSignature === lastToolCallSignature) {
+          repeatedToolCallCount++;
+          if (repeatedToolCallCount >= MAX_REPEATED_TOOL_CALLS) {
+            spinner.stop();
+            process.stderr.write(
+              theme.warning(`
+\u26A0 Detected ${repeatedToolCallCount + 1} identical consecutive tool calls \u2014 breaking loop.
+`)
+            );
+            extraMessages.push({
+              role: "user",
+              content: "You are stuck in a loop calling the same tool with the same arguments repeatedly. Stop calling tools and give the user a final text response summarizing what you found and what needs to be done next. " + (this.planMode ? "If you need to execute commands (bash, psql, etc.), tell the user to type `/plan execute` first." : "")
+            });
+            repeatedToolCallCount = 0;
+            lastToolCallSignature = "";
+            continue;
+          }
+        } else {
+          lastToolCallSignature = currentSignature;
+          repeatedToolCallCount = 0;
+        }
         if (this._userInterjection) {
           const msg = this._userInterjection;
           this._userInterjection = null;
@@ -5411,7 +5437,7 @@ program.command("web").description("Start Web UI server with browser-based chat
     console.error("Error: Invalid port number. Must be between 1 and 65535.");
     process.exit(1);
   }
-  const { startWebServer } = await import("./server-ZMMVUF7Q.js");
+  const { startWebServer } = await import("./server-HERHWMR4.js");
   await startWebServer({ port, host: options.host });
 });
 program.command("user [action] [username]").description("Manage Web UI users (list | create <name> | delete <name> | reset-password <name> | migrate <name>)").action(async (action, username) => {

package/dist/{run-tests-ITCKU6DW.js → run-tests-TGHLGADJ.js}

@@ -2,7 +2,7 @@
 import {
   executeTests,
   runTestsTool
-} from "./chunk-J6SXTNJO.js";
+} from "./chunk-NT7GYZEL.js";
 export {
   executeTests,
   runTestsTool

package/dist/{server-ZMMVUF7Q.js → server-HERHWMR4.js}

@@ -23,7 +23,7 @@ import {
   setupProxy,
   spawnAgentContext,
   truncateOutput
-} from "./chunk-UR5QRBEG.js";
+} from "./chunk-HLR3T3PP.js";
 import {
   AGENTIC_BEHAVIOR_GUIDELINE,
   CONTEXT_FILE_CANDIDATES,
@@ -35,7 +35,7 @@ import {
   PLAN_MODE_SYSTEM_ADDON,
   SKILLS_DIR_NAME,
   VERSION
-} from "./chunk-J6SXTNJO.js";
+} from "./chunk-NT7GYZEL.js";
 import {
   AuthManager
 } from "./chunk-CPLT6CD3.js";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jinzd-ai-cli",
-  "version": "0.2.9",
+  "version": "0.2.11",
   "description": "Cross-platform REPL-style AI CLI with multi-provider support",
   "type": "module",
   "main": "./dist/index.js",