jinzd-ai-cli 0.2.10 → 0.2.12
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +7 -2
- package/README.zh-CN.md +7 -2
- package/dist/{chunk-OOLIOMNF.js → chunk-GGAA3JRD.js} +1 -1
- package/dist/{chunk-MBIHQ34V.js → chunk-IGZDXYIX.js} +1 -1
- package/dist/index.js +44 -7
- package/dist/{run-tests-RAXI3LMY.js → run-tests-R2BHDFGB.js} +1 -1
- package/dist/{server-HYAIDK7B.js → server-Z3UV3EZB.js} +2 -2
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -9,6 +9,7 @@
|
|
|
9
9
|
[](https://nodejs.org)
|
|
10
10
|
[]()
|
|
11
11
|
[](https://github.com/jinzhengdong/ai-cli/releases)
|
|
12
|
+
[](https://github.com/jinzhengdong/ai-cli/actions/workflows/ci.yml)
|
|
12
13
|
|
|
13
14
|
**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.
|
|
14
15
|
|
|
@@ -26,13 +27,15 @@
|
|
|
26
27
|
- **Streaming Tool Use** — Real-time streaming of AI reasoning and tool calls as they happen
|
|
27
28
|
- **Sub-Agents** — Delegate complex subtasks to isolated child agents with independent tool loops
|
|
28
29
|
- **Extended Thinking** — Claude deep reasoning mode with `/think` toggle
|
|
29
|
-
- **Plan Mode** — Read-only planning phase (`/plan`) where AI analyzes before executing
|
|
30
|
+
- **Plan Mode** — Read-only planning phase (`/plan`) where AI analyzes before executing, with loop detection
|
|
31
|
+
- **Auto-Pause** — Automatically pauses every 10 rounds for user review and redirection
|
|
30
32
|
- **MCP Protocol** — Connect external MCP servers for dynamic tool discovery
|
|
31
33
|
- **Multi-User Auth** — Web UI supports multiple users with password authentication
|
|
32
34
|
- **PWA Support** — Install Web UI as a desktop/mobile app, accessible over LAN
|
|
33
35
|
- **Hierarchical Context** — 3-layer context files (global / project / subdirectory) auto-injected
|
|
34
36
|
- **Headless Mode** — `ai-cli -p "prompt"` for CI/CD pipelines and scripting
|
|
35
37
|
- **35+ REPL Commands** — Session management, checkpointing, code review, scaffolding, and more
|
|
38
|
+
- **GitHub Actions CI/CD** — Automated testing on Node 20/22 + npm publish on release tags
|
|
36
39
|
- **Cross-Platform** — Windows, macOS, Linux
|
|
37
40
|
|
|
38
41
|
## Installation
|
|
@@ -125,7 +128,7 @@ AI autonomously invokes these 16 tools during conversations:
|
|
|
125
128
|
| `bash` | varies | Execute shell commands (PowerShell on Windows, $SHELL on Unix) |
|
|
126
129
|
| `read_file` | safe | Read file contents (10 MB limit, image support) |
|
|
127
130
|
| `write_file` | write | Create/overwrite files (diff preview + confirmation) |
|
|
128
|
-
| `edit_file` | write | Precise string replacement with fuzzy matching hints |
|
|
131
|
+
| `edit_file` | write | Precise string replacement with fuzzy matching hints + `replaceAll` mode |
|
|
129
132
|
| `list_dir` | safe | List directory contents |
|
|
130
133
|
| `grep_files` | safe | Regex search across files |
|
|
131
134
|
| `glob_files` | safe | Match files by glob pattern |
|
|
@@ -167,6 +170,8 @@ AI autonomously invokes these 16 tools during conversations:
|
|
|
167
170
|
| `/config` | Open configuration wizard |
|
|
168
171
|
| `/help` | Show all available commands |
|
|
169
172
|
|
|
173
|
+
**Multi-line input**: Use `\` at end of line for continuation, or paste multi-line content directly (auto-detected and merged).
|
|
174
|
+
|
|
170
175
|
See the [full command reference](USAGE.md) for all 35+ commands.
|
|
171
176
|
|
|
172
177
|
## CLI Parameters
|
package/README.zh-CN.md
CHANGED
|
@@ -9,6 +9,7 @@
|
|
|
9
9
|
[](https://nodejs.org)
|
|
10
10
|
[]()
|
|
11
11
|
[](https://github.com/jinzhengdong/ai-cli/releases)
|
|
12
|
+
[](https://github.com/jinzhengdong/ai-cli/actions/workflows/ci.yml)
|
|
12
13
|
|
|
13
14
|
## 特性亮点
|
|
14
15
|
|
|
@@ -18,13 +19,15 @@
|
|
|
18
19
|
- **流式工具调用** — 实时流式展示 AI 推理过程和工具调用
|
|
19
20
|
- **子代理系统** — 将复杂子任务委派给独立子代理执行
|
|
20
21
|
- **深度推理** — Claude Extended Thinking,`/think` 一键切换
|
|
21
|
-
- **规划模式** — `/plan` 进入只读规划,AI
|
|
22
|
+
- **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
|
|
23
|
+
- **自动暂停** — 每 10 轮自动暂停,用户可审查进度或重定向 AI
|
|
22
24
|
- **MCP 协议** — 接入外部 MCP 服务器,动态发现工具
|
|
23
25
|
- **多用户认证** — Web UI 支持多用户密码登录
|
|
24
26
|
- **PWA 支持** — Web UI 可安装为桌面/移动应用,支持局域网访问
|
|
25
27
|
- **三层级上下文** — 全局 / 项目 / 子目录上下文文件自动注入
|
|
26
28
|
- **无头模式** — `aicli -p "提示词"` 用于 CI/CD 管道和脚本
|
|
27
29
|
- **35+ REPL 命令** — 会话管理、检查点、代码审查、脚手架等
|
|
30
|
+
- **GitHub Actions CI/CD** — Node 20/22 自动测试 + Release tag 自动发布 npm
|
|
28
31
|
- **跨平台** — Windows、macOS、Linux
|
|
29
32
|
|
|
30
33
|
## 安装
|
|
@@ -117,7 +120,7 @@ AI 在对话中可自主调用 16 个工具:
|
|
|
117
120
|
| `bash` | 动态判断 | 执行 shell 命令(Windows: PowerShell,Unix: $SHELL) |
|
|
118
121
|
| `read_file` | 安全 | 读取文件内容(10 MB 限制,支持图片) |
|
|
119
122
|
| `write_file` | 写入 | 创建/覆盖文件(diff 预览 + 确认) |
|
|
120
|
-
| `edit_file` | 写入 | 精确字符串替换,模糊匹配提示 |
|
|
123
|
+
| `edit_file` | 写入 | 精确字符串替换,模糊匹配提示 + `replaceAll` 全局替换模式 |
|
|
121
124
|
| `list_dir` | 安全 | 列出目录内容 |
|
|
122
125
|
| `grep_files` | 安全 | 正则搜索文件内容 |
|
|
123
126
|
| `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:
|
|
|
159
162
|
| `/config` | 打开配置向导 |
|
|
160
163
|
| `/help` | 显示所有命令 |
|
|
161
164
|
|
|
165
|
+
**多行输入**:行末加 `\` 续行,或直接粘贴多行内容(自动检测合并)。
|
|
166
|
+
|
|
162
167
|
完整命令参考见 [使用说明](USAGE.md)。
|
|
163
168
|
|
|
164
169
|
## CLI 参数
|
package/dist/index.js
CHANGED
|
@@ -35,7 +35,7 @@ import {
|
|
|
35
35
|
theme,
|
|
36
36
|
truncateOutput,
|
|
37
37
|
undoStack
|
|
38
|
-
} from "./chunk-
|
|
38
|
+
} from "./chunk-IGZDXYIX.js";
|
|
39
39
|
import {
|
|
40
40
|
AGENTIC_BEHAVIOR_GUIDELINE,
|
|
41
41
|
AUTHOR,
|
|
@@ -55,7 +55,7 @@ import {
|
|
|
55
55
|
REPO_URL,
|
|
56
56
|
SKILLS_DIR_NAME,
|
|
57
57
|
VERSION
|
|
58
|
-
} from "./chunk-
|
|
58
|
+
} from "./chunk-GGAA3JRD.js";
|
|
59
59
|
|
|
60
60
|
// src/index.ts
|
|
61
61
|
import { program } from "commander";
|
|
@@ -1904,7 +1904,7 @@ ${hint}` : "")
|
|
|
1904
1904
|
description: "Run project tests and show structured report",
|
|
1905
1905
|
usage: "/test [command|filter]",
|
|
1906
1906
|
async execute(args, _ctx) {
|
|
1907
|
-
const { executeTests } = await import("./run-tests-
|
|
1907
|
+
const { executeTests } = await import("./run-tests-R2BHDFGB.js");
|
|
1908
1908
|
const argStr = args.join(" ").trim();
|
|
1909
1909
|
let testArgs = {};
|
|
1910
1910
|
if (argStr) {
|
|
@@ -3785,7 +3785,15 @@ ${projectContext}`);
|
|
|
3785
3785
|
const envInfo = `OS: ${osName}
|
|
3786
3786
|
${shellInfo}
|
|
3787
3787
|
Working directory: ${process.cwd()}`;
|
|
3788
|
-
const
|
|
3788
|
+
const budgetInfo = `
|
|
3789
|
+
|
|
3790
|
+
# Tool Round Budget
|
|
3791
|
+
You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds per conversation turn. Plan your work accordingly:
|
|
3792
|
+
- Use efficient approaches: batch edits (replaceAll), read multiple files in one round, avoid re-reading files you already read.
|
|
3793
|
+
- On Windows, use PowerShell cmdlets (Invoke-RestMethod, Get-ChildItem) instead of Unix commands (curl, grep, find).
|
|
3794
|
+
- If starting a long-running server process via bash, use background execution \u2014 do not block the tool round.
|
|
3795
|
+
- Prioritize the most critical tasks first in case you run out of rounds.`;
|
|
3796
|
+
const parts = [dateTimeInfo + "\n" + envInfo, AGENTIC_BEHAVIOR_GUIDELINE + budgetInfo];
|
|
3789
3797
|
const memory = this.loadMemoryContent();
|
|
3790
3798
|
if (memory) {
|
|
3791
3799
|
parts.push(`# Persistent Memory
|
|
@@ -4821,6 +4829,7 @@ You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds for this task. Plan ef
|
|
|
4821
4829
|
let consecutiveFreeRounds = 0;
|
|
4822
4830
|
let lastToolCallSignature = "";
|
|
4823
4831
|
let repeatedToolCallCount = 0;
|
|
4832
|
+
const roundToolHistory = [];
|
|
4824
4833
|
this.setupInterjectionListener();
|
|
4825
4834
|
process.stdout.write(theme.dim(" (Type a message + Enter to redirect AI at any time)\n"));
|
|
4826
4835
|
try {
|
|
@@ -4841,12 +4850,24 @@ You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds for this task. Plan ef
|
|
|
4841
4850
|
extraMessages.push({ role: "user", content: cmd });
|
|
4842
4851
|
}
|
|
4843
4852
|
const roundsLeft = MAX_TOOL_ROUNDS - round;
|
|
4844
|
-
if (roundsLeft ===
|
|
4853
|
+
if (roundsLeft === 10) {
|
|
4854
|
+
extraMessages.push({
|
|
4855
|
+
role: "user",
|
|
4856
|
+
content: `\u{1F4CA} Budget note: ${roundsLeft} tool rounds remaining out of ${MAX_TOOL_ROUNDS}. Plan your remaining work efficiently \u2014 use batch operations (e.g., replaceAll) when possible.`
|
|
4857
|
+
});
|
|
4858
|
+
} else if (roundsLeft === 5) {
|
|
4845
4859
|
extraMessages.push({
|
|
4846
4860
|
role: "user",
|
|
4847
|
-
content: `\u26A0\uFE0F Budget warning: Only ${roundsLeft} tool rounds remaining. Prioritize completing the most critical task. If you cannot finish everything, summarize what's done and what remains.`
|
|
4861
|
+
content: `\u26A0\uFE0F Budget warning: Only ${roundsLeft} tool rounds remaining. Prioritize completing the most critical task. Use efficient approaches (batch edits, fewer reads). If you cannot finish everything, summarize what's done and what remains.`
|
|
4848
4862
|
});
|
|
4849
4863
|
process.stdout.write(theme.warning(` \u26A0\uFE0F Low budget: ${roundsLeft} rounds remaining
|
|
4864
|
+
`));
|
|
4865
|
+
} else if (roundsLeft === 3) {
|
|
4866
|
+
extraMessages.push({
|
|
4867
|
+
role: "user",
|
|
4868
|
+
content: `\u{1F6A8} Critical budget: Only ${roundsLeft} rounds left! Wrap up NOW \u2014 complete the current operation and give a final summary. Do NOT start new tasks.`
|
|
4869
|
+
});
|
|
4870
|
+
process.stdout.write(theme.error(` \u{1F6A8} Critical: ${roundsLeft} rounds remaining
|
|
4850
4871
|
`));
|
|
4851
4872
|
}
|
|
4852
4873
|
if (this._userInterjection) {
|
|
@@ -5031,6 +5052,10 @@ You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds for this task. Plan ef
|
|
|
5031
5052
|
process.stdin.removeListener("data", this._interjectionHandler);
|
|
5032
5053
|
}
|
|
5033
5054
|
const toolResults = await this.toolExecutor.executeAll(result.toolCalls);
|
|
5055
|
+
roundToolHistory.push({
|
|
5056
|
+
round: round + 1,
|
|
5057
|
+
tools: result.toolCalls.map((tc) => tc.name)
|
|
5058
|
+
});
|
|
5034
5059
|
if (this._interjectionHandler) {
|
|
5035
5060
|
process.stdin.on("data", this._interjectionHandler);
|
|
5036
5061
|
process.stdin.resume();
|
|
@@ -5083,6 +5108,18 @@ You have a maximum of ${MAX_TOOL_ROUNDS} tool call rounds for this task. Plan ef
|
|
|
5083
5108
|
process.stdout.write("\n");
|
|
5084
5109
|
process.stdout.write(theme.warning(`\u23F8 Auto-pause: ${effectiveRound}/${MAX_TOOL_ROUNDS} rounds used, ${remaining} remaining
|
|
5085
5110
|
`));
|
|
5111
|
+
const recentHistory = roundToolHistory.slice(-AUTO_PAUSE_INTERVAL);
|
|
5112
|
+
if (recentHistory.length > 0) {
|
|
5113
|
+
const toolCounts = /* @__PURE__ */ new Map();
|
|
5114
|
+
for (const rh of recentHistory) {
|
|
5115
|
+
for (const t of rh.tools) {
|
|
5116
|
+
toolCounts.set(t, (toolCounts.get(t) || 0) + 1);
|
|
5117
|
+
}
|
|
5118
|
+
}
|
|
5119
|
+
const summary = [...toolCounts.entries()].sort((a, b) => b[1] - a[1]).map(([name, count]) => count > 1 ? `${name}\xD7${count}` : name).join(", ");
|
|
5120
|
+
process.stdout.write(theme.dim(` Tools used: ${summary}
|
|
5121
|
+
`));
|
|
5122
|
+
}
|
|
5086
5123
|
process.stdout.write(theme.dim(" Press ") + theme.info("y") + theme.dim(" to continue, or ") + theme.info("type a message") + theme.dim(" to redirect AI:\n"));
|
|
5087
5124
|
process.stdout.write(theme.dim(" (Press ") + theme.info("n") + theme.dim(" or ") + theme.info("Esc") + theme.dim(" to stop)\n"));
|
|
5088
5125
|
this.teardownInterjectionListener();
|
|
@@ -5437,7 +5474,7 @@ program.command("web").description("Start Web UI server with browser-based chat
|
|
|
5437
5474
|
console.error("Error: Invalid port number. Must be between 1 and 65535.");
|
|
5438
5475
|
process.exit(1);
|
|
5439
5476
|
}
|
|
5440
|
-
const { startWebServer } = await import("./server-
|
|
5477
|
+
const { startWebServer } = await import("./server-Z3UV3EZB.js");
|
|
5441
5478
|
await startWebServer({ port, host: options.host });
|
|
5442
5479
|
});
|
|
5443
5480
|
program.command("user [action] [username]").description("Manage Web UI users (list | create <name> | delete <name> | reset-password <name> | migrate <name>)").action(async (action, username) => {
|
|
@@ -23,7 +23,7 @@ import {
|
|
|
23
23
|
setupProxy,
|
|
24
24
|
spawnAgentContext,
|
|
25
25
|
truncateOutput
|
|
26
|
-
} from "./chunk-
|
|
26
|
+
} from "./chunk-IGZDXYIX.js";
|
|
27
27
|
import {
|
|
28
28
|
AGENTIC_BEHAVIOR_GUIDELINE,
|
|
29
29
|
CONTEXT_FILE_CANDIDATES,
|
|
@@ -35,7 +35,7 @@ import {
|
|
|
35
35
|
PLAN_MODE_SYSTEM_ADDON,
|
|
36
36
|
SKILLS_DIR_NAME,
|
|
37
37
|
VERSION
|
|
38
|
-
} from "./chunk-
|
|
38
|
+
} from "./chunk-GGAA3JRD.js";
|
|
39
39
|
import {
|
|
40
40
|
AuthManager
|
|
41
41
|
} from "./chunk-CPLT6CD3.js";
|