min-agent 0.1.5 → 0.1.7

package/README.md

@@ -1,6 +1,6 @@
 # min-agent
 
-Minimal yet powerful AI coding agent for the terminal. Supports any OpenAI-compatible API, MCP servers, skills, memory, and multi-agent parallel execution.
+Minimal AI coding agent for the terminal. Works with any OpenAI-compatible API.
 
 ## Install
 
@@ -11,335 +11,134 @@ npm install -g min-agent
 ## Quick Start
 
 ```bash
-# 1. Configure your LLM provider
-min-agent setup
-
-# 2. Chat mode (general assistant)
-min-agent chat "what files are in this directory?"
-
-# 3. Code mode (project-aware AI coding)
-min-agent code
+min-agent setup              # Configure provider
+min-agent chat "hello"       # Single message
+min-agent chat               # Interactive session
+min-agent code               # Project-aware coding mode
 ```
 
 ---
 
 ## Commands
 
-### Chat
-
-```bash
-min-agent chat <message>          # Single-shot message
-min-agent chat                    # Interactive multi-turn session
-min-agent chat --resume <id>      # Resume a previous session
-min-agent chat -m <model> "msg"   # Use a specific model
-min-agent chat -i img.png "what's this?"  # Attach image
-```
-
-### Code (AI Coding Mode)
-
-```bash
-min-agent code                    # Project-aware coding session
-min-agent code --resume <id>      # Resume coding session
-```
-
-Code mode automatically:
-- Scans the project (language, framework, git status, structure)
-- Uses a coding-optimized system prompt
-- Includes the `explore` tool for deep codebase analysis
-- Follows project conventions (no comments, match existing style)
-
-### Configuration
-
 ```bash
-min-agent setup                   # Interactive provider setup
+min-agent chat [message]          # Chat (interactive if no message)
+min-agent chat -m <model> "msg"   # Override model
+min-agent chat -i img.png "msg"   # Attach image
+min-agent chat --resume <id>      # Resume session
+min-agent code                    # AI coding mode (project scan)
+min-agent setup                   # Configure API provider
 min-agent models                  # List available models
-min-agent rules                   # Show loaded instruction rules
-min-agent rules edit              # Edit global rules file
-```
-
-### Memory
-
-```bash
-min-agent memory                  # List all memories
-min-agent memory add "text"       # Add manually
-min-agent memory search "query"   # Search
-min-agent memory delete <n>       # Delete by number
-```
-
-The agent also saves memories automatically during conversations when you share preferences or important context.
-
-### Sessions
-
-```bash
 min-agent history                 # List saved sessions
-min-agent chat --resume <id>      # Resume a session
-```
-
-Sessions are auto-saved when you exit interactive chat/code mode, with LLM-generated titles.
-
-### MCP (Model Context Protocol)
-
-```bash
-min-agent mcp add <name> <cmd...>                    # Add local stdio server
-min-agent mcp add <name> --url <url> [--token <t>]   # Add remote HTTP server
-min-agent mcp remove <name>                          # Remove server
-min-agent mcp list                                   # List servers
-min-agent mcp check                                  # Check connectivity
-```
-
-### Skills
-
-```bash
-min-agent skills list             # List available skills
-```
-
-### HTTP API
-
-```bash
-min-agent serve                   # Start HTTP server (default :8787)
-min-agent serve --port 3000       # Custom port
+min-agent rules                   # Show loaded rules
+min-agent rules edit              # Edit global rules
+min-agent memory                  # List memories
+min-agent memory add "text"       # Save memory
+min-agent mcp -h                  # MCP server management
+min-agent skills -h               # Skills management
+min-agent serve                   # Start HTTP API server
 ```
 
-See [docs/API.md](docs/API.md) for full API reference.
-
----
-
 ## Interactive Commands
 
-In chat/code mode, use slash commands:
-
 | Command | Description |
 |---------|-------------|
-| `/clear` | Clear conversation history |
-| `/compact` | Force context compaction |
-| `/model [name]` | Show or change model |
-| `/memory [text]` | List or save memory |
-| `/tokens` | Show context window usage |
-| `/path` | Show working directory |
-| `/help` | Show all commands |
-| `/exit` | Exit session |
-
-**Ctrl+C** interrupts the current generation without exiting.
+| `/clear` | Clear history |
+| `/compact` | Force compaction |
+| `/model [name]` | Show/change model |
+| `/memory [text]` | List/save memory |
+| `/tokens` | Context usage |
+| `/paste [text]` | Clipboard image |
+| `/path` | Working directory |
+| `/help` | All commands |
+| `/exit` | Exit |
+
+**Ctrl+C** interrupts current generation.
 
 ---
 
 ## Configuration
 
-All config lives in `~/.min-agent/`:
-
-```
-~/.min-agent/
-├── config.json              # Provider, model, instructions
-├── rules.md                 # Global rules (always injected)
-├── mcp.json                 # MCP server definitions
-├── memory.json              # Persistent memories
-├── models-cache.json        # Model list cache (fallback)
-├── context-window-cache.json # Context window cache
-└── sessions/                # Saved conversation sessions
-```
-
-### config.json
+All config in `~/.min-agent/`:
 
 ```json
+// ~/.min-agent/config.json
 {
   "provider": {
     "type": "openai-compatible",
     "baseURL": "https://api.openai.com/v1",
     "apiKey": "sk-...",
-    "defaultModel": "gpt-4o",
+    "defaultModel": "gpt-5.5",
     "contextWindow": 128000
   },
-  "instructions": [
-    "./docs/CODING_RULES.md",
-    "https://example.com/team-rules.md"
-  ]
+  "instructions": ["./docs/rules.md"],
+  "disabledSkills": []
 }
 ```
 
-### Provider Types
-
-| Type | Description |
-|------|-------------|
-| `openai-compatible` | Any OpenAI-compatible API (MiniMax, DeepSeek, vLLM, etc.) |
-| `openai` | OpenAI official |
-| `ollama` | Local Ollama (`http://localhost:11434/v1`) |
-
-### Context Window Auto-Detection
-
-The context window size is automatically detected from:
-1. `config.provider.contextWindow` (manual override)
-2. OpenRouter API (`context_length`)
-3. Ollama API (`/api/show`)
-4. vLLM API (`max_model_len`)
-5. [models.dev](https://models.dev) database
-6. Fallback: 128,000 tokens
+Provider types: `openai-compatible` | `openai` | `ollama`
 
 ---
 
-## Tools
-
-### Built-in
-
-| Tool | Description |
-|------|-------------|
-| `bash` | Shell commands (async, timeout, Ctrl+C killable) |
-| `read` | Read files (with context-aware instruction discovery) |
-| `write` | Write/create files (with overwrite confirmation) |
-| `edit` | Search/replace precise editing |
-| `glob` | Find files by pattern |
-| `grep` | Search file contents with regex |
-| `web_search` | Web search via SearXNG |
-| `web_fetch` | Fetch URLs (with Firecrawl fallback for SPAs) |
-| `todo` | Task tracking with progress visibility |
-| `question` | Ask user for clarification |
-| `task` | Spawn parallel sub-agents |
-| `explore` | Deep codebase exploration (read-only sub-agent) |
-| `skill` | Load domain-specific instructions |
-| `memory_save` | Save information for future sessions |
-| `memory_search` | Search saved memories |
-| `memory_delete` | Remove a memory |
-| `codesearch` | Semantic code search via Exa (requires `EXA_API_KEY`) |
-
-### Sub-Agent System
-
-The `task` tool spawns independent sub-agents that:
-- Have their own context and tool access
-- Can run in parallel (call multiple tasks at once)
-- Are limited to 15 steps with doom loop detection
-- Cannot spawn further sub-agents (no recursion)
-
-The `explore` tool is a specialized read-only sub-agent for codebase exploration with configurable thoroughness (quick/medium/thorough).
-
-### Plugins
-
-Custom tools can be loaded from:
-- `.min-agent/tools/*.ts` (project-level)
-- `~/.min-agent/tools/*.ts` (global)
-
-```typescript
-// .min-agent/tools/deploy.ts
-export const deploy = {
-  description: "Deploy the application",
-  parameters: { env: { type: "string", description: "Target environment" } },
-  execute: async ({ env }) => {
-    // your logic
-    return `Deployed to ${env}`
-  }
-}
-```
-
----
-
-## MCP (Model Context Protocol)
-
-Connect external tool servers:
+## MCP
 
 ```bash
-# Local stdio servers
-min-agent mcp add filesystem npx -y @modelcontextprotocol/server-filesystem /tmp
-min-agent mcp add fetch npx -y @modelcontextprotocol/server-fetch
-
-# Remote HTTP servers
-min-agent mcp add remote-api --url https://mcp.example.com --token $TOKEN
+min-agent mcp add <name> <cmd...>              # Local stdio
+min-agent mcp add <name> --url <url> --token T # Remote HTTP
+min-agent mcp list                             # List (name + status)
+min-agent mcp info <name>                      # Details
+min-agent mcp enable/disable <name...>         # Toggle
+min-agent mcp remove <name>                    # Remove
+min-agent mcp check                            # Test connectivity
 ```
 
-MCP tools are automatically namespaced (`servername_toolname`) and available in all modes.
-
-Supports:
-- Stdio transport (local processes)
-- Streamable HTTP transport
-- SSE transport (legacy)
-- OAuth authentication detection
-- Auto-reconnect on failure
-
----
-
 ## Skills
 
-Skills provide domain-specific instructions loaded on demand.
-
-```
-.min-agent/skills/my-skill/SKILL.md
+```bash
+min-agent skills list                          # List (name + status)
+min-agent skills info <name>                   # Details
+min-agent skills enable/disable <name...>      # Toggle
 ```
 
+Create: `.min-agent/skills/<name>/SKILL.md` or `~/.agents/skills/<name>/SKILL.md`
+
 ```markdown
 ---
 name: my-skill
-description: What this skill does
+description: What it does
 ---
-
-# Instructions for the agent...
+# Instructions...
 ```
 
-Discovery locations:
-- `.min-agent/skills/` (project)
-- `.opencode/skills/` (project)
-- `.claude/skills/` (project, compatibility)
-
----
-
-## Rules (Instructions)
-
-Rules are injected into every conversation as system instructions.
-
-| Source | Priority |
-|--------|----------|
-| `~/.min-agent/rules.md` | Global (always loaded) |
-| `./AGENTS.md` or `./RULES.md` | Project-level |
-| `.min-agent/AGENTS.md` | Project-level |
-| `config.instructions[]` | Paths or URLs |
+## Rules
 
-**Context-aware**: When the agent reads a file, nearby `AGENTS.md`/`RULES.md` files are automatically discovered and injected.
+| Source | Scope |
+|--------|-------|
+| `~/.min-agent/rules.md` | Global |
+| `./AGENTS.md` or `./RULES.md` | Project |
+| `config.instructions[]` | Custom paths/URLs |
 
----
-
-## Memory
+## Plugins
 
-The agent remembers things across sessions:
+Custom tools: `.min-agent/tools/*.ts` or `~/.min-agent/tools/*.ts`
 
-```bash
-min-agent memory add "I prefer functional style"
-min-agent memory add "Project uses PostgreSQL"
+```typescript
+export const myTool = {
+  description: "...",
+  parameters: { arg: { type: "string", description: "..." } },
+  execute: async ({ arg }) => "result"
+}
 ```
 
-Or just tell the agent: "记住：我喜欢用中文回复" — it will automatically save the memory.
-
-Memories are injected into the system prompt so the agent always has access to them.
-
----
-
-## Safety Features
-
-- **Dangerous command confirmation**: `rm -rf`, `sudo`, `git push --force` etc. require approval
-- **File overwrite confirmation**: Existing files prompt before overwriting
-- **Doom loop detection**: Breaks infinite tool call loops (3 identical calls)
-- **Auto-compaction**: Prevents context overflow with structured summarization
-- **`--yes` / `-y` flag**: Skip all confirmations (for automation)
-
----
-
-## Context Management
-
-- **Real token tracking**: Uses API response usage data (not estimation)
-- **Auto-compaction**: Triggers at 75% of context window
-- **Structured summaries**: Goal/Progress/Decisions/Files format
-- **Incremental updates**: Updates existing summary instead of rewriting
-- **Tool output pruning**: Trims old tool outputs to save space
-- **Auto-continue**: After compaction, agent continues working automatically
-
 ---
 
 ## HTTP API
 
-Start with `min-agent serve`. Full reference: [docs/API.md](docs/API.md)
+```bash
+min-agent serve [--port 8787]
+```
 
-Key endpoints:
-- `POST /v1/chat` — Chat (streaming SSE or JSON)
-- `GET /v1/sessions` — List sessions
-- `GET /v1/memory` — List memories
-- `GET /v1/mcp` — MCP status
-- `GET /v1/project` — Project scan
-- `POST /v1/chat/compact` — Manual compaction
+Full reference: [docs/API.md](docs/API.md)
 
 ---
 
@@ -347,13 +146,10 @@ Key endpoints:
 
 | Variable | Description |
 |----------|-------------|
-| `AGENT_MODEL` | Override default model |
-| `EXA_API_KEY` | Enable Exa code search tool |
-| `MIN_AGENT_SHOW_THINKING` | Set to `0` to hide thinking output |
-| `MIN_AGENT_SERVE_TOKEN` | API auth token for serve mode |
-| `MIN_AGENT_SERVE_CORS` | Enable CORS for serve mode |
-
----
+| `EXA_API_KEY` | Enable Exa code search |
+| `MIN_AGENT_SHOW_THINKING` | `0` to hide thinking |
+| `MIN_AGENT_SERVE_TOKEN` | API auth token |
+| `MIN_AGENT_SERVE_CORS` | `1` to enable CORS |
 
 ## License
 

package/dist/agent.js

@@ -2,7 +2,7 @@ import { streamText, stepCountIs } from "ai";
 import { readFileSync, existsSync } from "fs";
 import path from "path";
 import { resolveModel } from "./provider.js";
-import { createTools } from "./tools/index.js";
+import { createChatTools, createCodeTools } from "./tools/index.js";
 import { initMcp, shutdownMcp, getMcpTools, loadMcpConfig, getMcpStatus } from "./mcp.js";
 import { discoverSkills, getSkillsTool, getSkillsSystemPrompt, getSkills } from "./skills.js";
 import { loadInstructions } from "./instructions.js";
@@ -53,7 +53,7 @@ function buildSystemPrompt(instructions) {
         `Platform: ${process.platform}`,
         `Date: ${new Date().toDateString()}`,
         "",
-        "min-agent: when the user asks to configure or install this agent (MCP, skills, rules, CLI, HTTP API, etc.), use the **read** tool on `README.md` in the working directory first, then follow what it says.",
+        `When the user asks to configure, install, or manage MCP servers, skills, rules, memory, or other min-agent features, use the read tool on the file at ${path.resolve(path.dirname(new URL(import.meta.url).pathname), "../README.md")} first, then follow what it says.`,
     ];
     const skillsPrompt = getSkillsSystemPrompt();
     if (skillsPrompt) {
@@ -159,45 +159,63 @@ export async function runChat(modelId, resumeSessionId) {
     });
     console.log("\x1b[90m输入消息开始对话，输入 /help 查看命令，/exit 退出\x1b[0m\n");
     rl.prompt();
-    for await (const line of rl) {
-        const input = line.trim();
+    // Multi-line paste detection: collect rapid successive lines
+    let pasteBuffer = [];
+    let pasteTimer = null;
+    const PASTE_DEBOUNCE_MS = 50;
+    const processInput = async (text) => {
+        const input = text.trim();
         if (!input) {
             rl.prompt();
-            continue;
+            return;
         }
-        // Handle slash commands
         if (input.startsWith("/")) {
             const handled = await handleSlashCommand(input, messages, instructions, modelId, rl, tracker);
-            if (handled === "exit")
-                break;
+            if (handled === "exit") {
+                rl.close();
+                return;
+            }
             if (handled === "paste") {
                 console.log();
                 abortController = new AbortController();
                 await runOnce(messages, instructions, modelId, abortController.signal, undefined, tracker);
                 abortController = null;
                 console.log();
-                rl.prompt();
-                continue;
             }
             rl.prompt();
-            continue;
+            return;
         }
+        // Show paste feedback for large inputs
+        const { processPastedInput, printPasteFeedback } = await import("./paste-handler.js");
+        const pasteResult = processPastedInput(input);
+        printPasteFeedback(pasteResult);
         console.log();
-        messages.push({ role: "user", content: input });
+        messages.push({ role: "user", content: pasteResult.fullText });
         abortController = new AbortController();
         await runOnce(messages, instructions, modelId, abortController.signal, undefined, tracker);
         abortController = null;
         console.log();
         rl.prompt();
-    }
-    // Auto-save session on exit with LLM-generated title
+    };
+    rl.on("line", (line) => {
+        pasteBuffer.push(line);
+        if (pasteTimer)
+            clearTimeout(pasteTimer);
+        pasteTimer = setTimeout(() => {
+            const combined = pasteBuffer.join("\n");
+            pasteBuffer = [];
+            pasteTimer = null;
+            processInput(combined);
+        }, PASTE_DEBOUNCE_MS);
+    });
+    // Wait for close
+    await new Promise((resolve) => rl.on("close", resolve));
+    // Auto-save session on exit
     if (messages.length > 0) {
-        const { saveSessionWithTitle } = await import("./sessions.js");
-        const model = resolveModel(modelId);
-        sessionId = await saveSessionWithTitle(messages, model, sessionId);
+        const { saveSession } = await import("./sessions.js");
+        sessionId = saveSession(messages, sessionId);
         console.log(`\x1b[90m  Session saved: ${sessionId}\x1b[0m`);
     }
-    rl.close();
     printDivider();
     console.log("\x1b[90mBye!\x1b[0m");
     await shutdownMcp();
@@ -248,41 +266,57 @@ export async function runCode(modelId, resumeSessionId) {
     });
     console.log("\x1b[90m输入任务开始编码，/help 查看命令，Ctrl+C 中断\x1b[0m\n");
     rl.prompt();
-    for await (const line of rl) {
-        const input = line.trim();
+    let pasteBuffer = [];
+    let pasteTimer = null;
+    const PASTE_DEBOUNCE_MS = 50;
+    const processCodeInput = async (text) => {
+        const input = text.trim();
         if (!input) {
             rl.prompt();
-            continue;
+            return;
         }
         if (input.startsWith("/")) {
             const handled = await handleSlashCommand(input, messages, [codePrompt], modelId, rl, tracker);
-            if (handled === "exit")
-                break;
+            if (handled === "exit") {
+                rl.close();
+                return;
+            }
             if (handled === "paste") {
                 console.log();
                 abortController = new AbortController();
                 await runOnceWithSystem(messages, codePrompt, modelId, abortController.signal, undefined, tracker);
                 abortController = null;
                 console.log();
-                rl.prompt();
-                continue;
             }
             rl.prompt();
-            continue;
+            return;
         }
+        const { processPastedInput, printPasteFeedback } = await import("./paste-handler.js");
+        const pasteResult = processPastedInput(input);
+        printPasteFeedback(pasteResult);
         console.log();
-        messages.push({ role: "user", content: input });
+        messages.push({ role: "user", content: pasteResult.fullText });
         abortController = new AbortController();
-        // Use code system prompt instead of generic one
         await runOnceWithSystem(messages, codePrompt, modelId, abortController.signal, undefined, tracker);
         abortController = null;
         console.log();
         rl.prompt();
-    }
+    };
+    rl.on("line", (line) => {
+        pasteBuffer.push(line);
+        if (pasteTimer)
+            clearTimeout(pasteTimer);
+        pasteTimer = setTimeout(() => {
+            const combined = pasteBuffer.join("\n");
+            pasteBuffer = [];
+            pasteTimer = null;
+            processCodeInput(combined);
+        }, PASTE_DEBOUNCE_MS);
+    });
+    await new Promise((resolve) => rl.on("close", resolve));
     if (messages.length > 0) {
-        const { saveSessionWithTitle } = await import("./sessions.js");
-        const model = resolveModel(modelId);
-        sessionId = await saveSessionWithTitle(messages, model, sessionId);
+        const { saveSession } = await import("./sessions.js");
+        sessionId = saveSession(messages, sessionId);
         console.log(`\x1b[90m  Session saved: ${sessionId}\x1b[0m`);
     }
     rl.close();
@@ -308,7 +342,7 @@ export async function runOnceWithSystem(messages, systemPrompt, modelId, abortSi
             console.log(`\x1b[90m  ✓ Compacted (${estimateTokens(messages)} tokens estimated)\x1b[0m`);
         }
     }
-    const builtinTools = createTools();
+    const builtinTools = createCodeTools();
     const mcpTools = getMcpTools();
     const memoryTools = getMemoryTools();
     const pluginTools = await loadPluginTools();
@@ -499,6 +533,7 @@ async function handleSlashCommand(input, messages, instructions, modelId, rl, tr
         case "exit":
         case "quit":
         case "q":
+            console.log("\x1b[90m⟳ Exiting...\x1b[0m");
             return "exit";
         case "clear":
             messages.length = 0;
@@ -581,12 +616,11 @@ async function handleSlashCommand(input, messages, instructions, modelId, rl, tr
             discoverSkills();
             const skills = getSkills();
             if (skills.length === 0) {
-                console.log("\x1b[90m  No skills found\x1b[0m");
+                console.log("\x1b[90m  No skills available\x1b[0m");
             }
             else {
-                console.log(`\x1b[90m  Available skills (${skills.length}):\x1b[0m`);
                 for (const skill of skills) {
-                    console.log(`\x1b[90m  - ${skill.name}: ${skill.description}\x1b[0m`);
+                    console.log(`\x1b[90m  ${skill.name}  enabled\x1b[0m`);
                 }
             }
             return "handled";
@@ -698,12 +732,11 @@ export async function runOnce(messages, instructions, modelId, abortSignal, call
             }
         }
     }
-    // Merge all tools: builtin + MCP + skill + memory + plugins + task
-    const builtinTools = createTools();
+    // Merge tools: chat builtin + MCP + skill + memory + plugins (no task/explore in chat mode)
+    const builtinTools = createChatTools();
     const mcpTools = getMcpTools();
     const memoryTools = getMemoryTools();
     const pluginTools = await loadPluginTools();
-    const { createTaskTool } = await import("./tools/task.js");
     const skills = getSkills();
     const allTools = { ...builtinTools, ...memoryTools, ...pluginTools };
     for (const [id, t] of Object.entries(mcpTools)) {
@@ -712,7 +745,6 @@ export async function runOnce(messages, instructions, modelId, abortSignal, call
     if (skills.length > 0) {
         allTools["skill"] = getSkillsTool();
     }
-    allTools["task"] = createTaskTool(modelId);
     let stepCount = 0;
     let hasError = false;
     const doomLoop = new DoomLoopDetector();