miii-agent 0.1.25 → 0.1.26

package/README.md

@@ -8,6 +8,7 @@
 
 <p align="center">
   <a href="https://www.npmjs.com/package/miii-agent"><img src="https://img.shields.io/npm/v/miii-agent" alt="npm version"></a>
+  <a href="https://www.npmjs.com/package/miii-agent"><img src="https://img.shields.io/npm/dt/miii-agent" alt="npm total downloads"></a>
   <a href="LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue.svg" alt="license"></a>
   <a href="https://nodejs.org"><img src="https://img.shields.io/badge/node-%3E%3D18-brightgreen" alt="node version"></a>
   <a href="https://ollama.com"><img src="https://img.shields.io/badge/powered%20by-Ollama-black" alt="powered by Ollama"></a>
@@ -68,6 +69,10 @@ It doesn't just chat, either — it decomposes the problem, calls tools, and che
   miii doctor                  # grade every installed model
   miii doctor qwen2.5-coder:7b # grade one
   ```
+- **🖼️ Paste images** — copy a screenshot and hit `Ctrl+V` to attach it to your message, or paste an image file path. Great for "why does this UI look broken?" or reading an error screenshot. **Needs a vision-capable model** (`llava`, `llama3.2-vision`, `qwen2-vl`, …) — text-only models silently ignore the image.
+  ```bash
+  ollama pull llava            # or llama3.2-vision
+  ```
 - **💧 Lossless output spill** — that 50K-line test log won't get truncated and leave the model guessing. miii spills the full output to disk and lets the model page through it. Nothing is ever lost.
 - **🔒 Permission-gated tools** — you approve what the agent can touch; "always" approvals persist. File tools are confined to your working directory.
 - **📄 `MIII.md`** — drop one in your repo to teach miii your conventions, build/test commands, and do's & don'ts. Same idea as `CLAUDE.md`, read every turn.
@@ -96,6 +101,7 @@ File tools (`read_file`, `write_file`, `edit_file`) reject `../` traversal and a
 |-----|--------|
 | `Enter` | Send prompt |
 | `@filename` | Attach file to context |
+| `Ctrl+V` | Paste clipboard image (needs a vision model) |
 | `/models` | Switch active model |
 | `/clear` | Reset conversation |
 | `Esc` | Stop generation or tool run |

package/dist/cli.js

@@ -129,6 +129,15 @@ function isConnectionError(err) {
   const msg = err instanceof Error ? err.message : String(err);
   return msg.includes("ECONNREFUSED") || msg.includes("fetch failed") || msg.includes("connect");
 }
+function isNoVisionError(err) {
+  const msg = (err instanceof Error ? err.message : String(err)).toLowerCase();
+  return msg.includes("does not support image") || msg.includes("image input");
+}
+function noVisionError(model) {
+  return new Error(
+    `"${model}" can't read images \u2014 it has no vision support. Switch to a vision-capable model (e.g. llava, llama3.2-vision, qwen2-vl) with /models, then resend. Pull one with: ollama pull llava`
+  );
+}
 async function listModels(entry) {
   try {
     const { models } = await makeClient(entry).list();
@@ -208,6 +217,7 @@ async function* chat(entry, model, messages, tools, opts) {
     }
   } catch (err) {
     if (signal?.aborted) return;
+    if (isNoVisionError(err)) throw noVisionError(model);
     if (isConnectionError(err)) {
       throw new Error(NOT_RUNNING);
     }
@@ -233,6 +243,7 @@ async function* chat(entry, model, messages, tools, opts) {
     }
   } catch (err) {
     if (opts?.signal?.aborted) return;
+    if (isNoVisionError(err)) throw noVisionError(model);
     if (isConnectionError(err)) {
       throw new Error(NOT_RUNNING);
     }
@@ -1429,7 +1440,9 @@ function toOllamaMessages(history, system) {
   const out = [{ role: "system", content: system }];
   for (const msg of history) {
     if (typeof msg.content === "string") {
-      out.push({ role: msg.role === "system" ? "system" : msg.role, content: msg.content });
+      const om = { role: msg.role === "system" ? "system" : msg.role, content: msg.content };
+      if (msg.role === "user" && msg.images && msg.images.length > 0) om.images = msg.images;
+      out.push(om);
       continue;
     }
     if (msg.role === "assistant") {
@@ -1615,7 +1628,11 @@ async function* runAgent(opts) {
   const effort = EFFORT_OPTIONS[loadConfig().effort ?? "medium"];
   const history = [
     ...opts.history,
-    { role: "user", content: opts.userText }
+    {
+      role: "user",
+      content: opts.userText,
+      ...opts.images && opts.images.length > 0 ? { images: opts.images } : {}
+    }
   ];
   let promptTokens = 0;
   let evalTokens = 0;
@@ -1833,14 +1850,14 @@ var init_loop = __esm({
 });
 
 // eval/runner.ts
-import { mkdtempSync, writeFileSync as writeFileSync7, mkdirSync as mkdirSync6, rmSync as rmSync3 } from "fs";
-import { dirname as dirname3, join as join8 } from "path";
-import { tmpdir } from "os";
+import { mkdtempSync, writeFileSync as writeFileSync7, mkdirSync as mkdirSync6, rmSync as rmSync4 } from "fs";
+import { dirname as dirname3, join as join10 } from "path";
+import { tmpdir as tmpdir2 } from "os";
 async function runScenario(model, s) {
-  const dir = mkdtempSync(join8(tmpdir(), "miii-eval-"));
+  const dir = mkdtempSync(join10(tmpdir2(), "miii-eval-"));
   const prevCwd = process.cwd();
   for (const [rel, content] of Object.entries(s.files ?? {})) {
-    const abs = join8(dir, rel);
+    const abs = join10(dir, rel);
     mkdirSync6(dirname3(abs), { recursive: true });
     writeFileSync7(abs, content, "utf-8");
   }
@@ -1880,7 +1897,7 @@ async function runScenario(model, s) {
   r.durationMs = Date.now() - start;
   if (r.error) {
     r.reason = `loop error: ${r.error}`;
-    rmSync3(dir, { recursive: true, force: true });
+    rmSync4(dir, { recursive: true, force: true });
     return r;
   }
   try {
@@ -1890,7 +1907,7 @@ async function runScenario(model, s) {
   } catch (err) {
     r.reason = `check threw: ${err instanceof Error ? err.message : String(err)}`;
   }
-  rmSync3(dir, { recursive: true, force: true });
+  rmSync4(dir, { recursive: true, force: true });
   return r;
 }
 var autoYes;
@@ -1903,13 +1920,13 @@ var init_runner = __esm({
 });
 
 // eval/scenarios.ts
-import { readFileSync as readFileSync7, existsSync as existsSync6 } from "fs";
-import { join as join9 } from "path";
+import { readFileSync as readFileSync9, existsSync as existsSync8 } from "fs";
+import { join as join11 } from "path";
 var read, scenarios;
 var init_scenarios = __esm({
   "eval/scenarios.ts"() {
     "use strict";
-    read = (dir, f) => existsSync6(join9(dir, f)) ? readFileSync7(join9(dir, f), "utf-8") : null;
+    read = (dir, f) => existsSync8(join11(dir, f)) ? readFileSync9(join11(dir, f), "utf-8") : null;
     scenarios = [
       {
         name: "edit-exact-string",
@@ -2059,7 +2076,7 @@ init_client();
 init_config();
 import { useState as useState5, useEffect as useEffect4, useRef as useRef2 } from "react";
 import { Box as Box13, Text as Text13, useApp } from "ink";
-import { homedir as homedir6 } from "os";
+import { homedir as homedir7 } from "os";
 import { sep as sep2 } from "path";
 
 // src/ui/WelcomeBlock.tsx
@@ -2346,7 +2363,8 @@ function persistSession(id, messages, title) {
   };
   const lines = [JSON.stringify({ type: "meta", ...meta })];
   for (const message of messages) {
-    lines.push(JSON.stringify({ type: "message", message }));
+    const { images: _img, ...rest } = message;
+    lines.push(JSON.stringify({ type: "message", message: rest }));
   }
   writeFileSync2(sessionPath(id), lines.join("\n") + "\n", "utf-8");
 }
@@ -3240,6 +3258,7 @@ var EMPTY_STATE_HINTS = [
   "\u2022 /models \u2014 switch model or effort",
   "\u2022 /new \u2014 start a new chat",
   "\u2022 /sessions \u2014 view saved chats",
+  "\u2022 ctrl+v \u2014 paste an image (needs a vision model)",
   "\u2022 ctrl+t \u2014 toggle thinking"
 ];
 var EMPTY_STATE_TITLE = "Ask anything, or try:";
@@ -3685,7 +3704,7 @@ function useAgentRunner(model, activeCtx) {
     setPendingPermission(null);
     req.resolve(answers[cursor]);
   }
-  async function sendMessage(text) {
+  async function sendMessage(text, images) {
     if (busyRef.current || !model) return;
     busyRef.current = true;
     setBusy(true);
@@ -3743,6 +3762,7 @@ function useAgentRunner(model, activeCtx) {
         cwd: process.cwd(),
         history: agentHistory,
         userText: text,
+        images,
         permissions: { ask: askPermission },
         signal: controller.signal,
         num_ctx: activeCtx ?? void 0
@@ -3872,16 +3892,141 @@ function useAgentRunner(model, activeCtx) {
 }
 
 // src/ui/hooks/useKeyboard.ts
-init_config();
+import { existsSync as existsSync7, readFileSync as readFileSync8 } from "fs";
+import { basename, join as join9 } from "path";
+import { homedir as homedir6 } from "os";
 import { useInput, useStdout } from "ink";
+
+// src/ui/clipboard.ts
+import { execFileSync as execFileSync2 } from "child_process";
+import { existsSync as existsSync6, readFileSync as readFileSync7, rmSync as rmSync3 } from "fs";
+import { join as join8 } from "path";
+import { tmpdir } from "os";
+function safeRm(p) {
+  try {
+    rmSync3(p, { force: true });
+  } catch {
+  }
+}
+function consume(p) {
+  try {
+    const b64 = readFileSync7(p).toString("base64");
+    return b64.length > 0 ? b64 : null;
+  } catch {
+    return null;
+  } finally {
+    safeRm(p);
+  }
+}
+function readMac(out) {
+  try {
+    execFileSync2("pngpaste", [out], { stdio: "ignore" });
+    if (existsSync6(out)) return consume(out);
+  } catch {
+  }
+  const png = "\xABclass PNGf\xBB";
+  const script = [
+    "try",
+    `set f to open for access (POSIX file "${out}") with write permission`,
+    `set theData to (the clipboard as ${png})`,
+    "write theData to f",
+    "close access f",
+    "on error",
+    "try",
+    "close access f",
+    "end try",
+    'return "NOIMG"',
+    "end try"
+  ];
+  try {
+    const res = execFileSync2("osascript", script.flatMap((s) => ["-e", s]), { encoding: "utf8" });
+    if (res.includes("NOIMG")) {
+      safeRm(out);
+      return null;
+    }
+    if (existsSync6(out)) return consume(out);
+  } catch {
+  }
+  safeRm(out);
+  return null;
+}
+function readLinux(out) {
+  for (const [cmd2, args2] of [
+    ["wl-paste", ["--type", "image/png"]],
+    ["xclip", ["-selection", "clipboard", "-t", "image/png", "-o"]]
+  ]) {
+    try {
+      const buf = execFileSync2(cmd2, args2, { maxBuffer: 64 * 1024 * 1024 });
+      if (buf.length > 0) return buf.toString("base64");
+    } catch {
+    }
+  }
+  return null;
+}
+function readWindows(out) {
+  const ps = [
+    "Add-Type -AssemblyName System.Windows.Forms,System.Drawing;",
+    "$img = [System.Windows.Forms.Clipboard]::GetImage();",
+    // Single-quoted PS string → backslashes are literal, no escaping needed.
+    `if ($img -ne $null) { $img.Save('${out}', [System.Drawing.Imaging.ImageFormat]::Png); 'OK' } else { 'NOIMG' }`
+  ].join(" ");
+  try {
+    const res = execFileSync2(
+      "powershell",
+      ["-NoProfile", "-NonInteractive", "-STA", "-Command", ps],
+      { encoding: "utf8" }
+    );
+    if (res.includes("NOIMG")) {
+      safeRm(out);
+      return null;
+    }
+    if (existsSync6(out)) return consume(out);
+  } catch {
+  }
+  safeRm(out);
+  return null;
+}
+function readClipboardImage() {
+  const out = join8(tmpdir(), `miii-clip-${Date.now()}-${Math.random().toString(36).slice(2)}.png`);
+  if (process.platform === "darwin") return readMac(out);
+  if (process.platform === "linux") return readLinux(out);
+  if (process.platform === "win32") return readWindows(out);
+  return null;
+}
+
+// src/ui/hooks/useKeyboard.ts
+init_config();
 var EFFORTS = ["low", "medium", "high"];
 var PASTE_CHIP_LINES = 4;
 var PASTE_CHIP_CHARS = 200;
 var pasteStore = /* @__PURE__ */ new Map();
 var pasteCounter = 0;
+var imageStore = /* @__PURE__ */ new Map();
+var imageCounter = 0;
+var IMAGE_EXT_RE = /\.(png|jpe?g|webp|gif|bmp)$/i;
 function clearPasteStore() {
   pasteStore.clear();
   pasteCounter = 0;
+  imageStore.clear();
+  imageCounter = 0;
+}
+function tryImagePaste(cleaned) {
+  let p = cleaned.trim();
+  if (p.startsWith('"') && p.endsWith('"') || p.startsWith("'") && p.endsWith("'")) {
+    p = p.slice(1, -1);
+  }
+  p = p.replace(/\\ /g, " ");
+  if (p.includes("\n") || !IMAGE_EXT_RE.test(p)) return null;
+  if (p.startsWith("~/")) p = join9(homedir6(), p.slice(2));
+  if (!existsSync7(p)) return null;
+  try {
+    const b64 = readFileSync8(p).toString("base64");
+    const chip = `[Image #${++imageCounter} \xB7 ${basename(p)}]`;
+    imageStore.set(chip, b64);
+    return chip;
+  } catch {
+    return null;
+  }
 }
 var inputHistory = [];
 var historyIndex = -1;
@@ -3902,6 +4047,8 @@ function stripControls(chunk) {
 function sanitizePaste(chunk) {
   if (chunk.length <= 1) return chunk;
   const cleaned = stripControls(chunk).replace(/\r/g, "");
+  const imageChip = tryImagePaste(cleaned);
+  if (imageChip) return imageChip;
   const lines = cleaned.split("\n").length;
   if (lines > PASTE_CHIP_LINES || cleaned.length > PASTE_CHIP_CHARS) {
     const chip = `[Pasted #${++pasteCounter} \xB7 ${lines} line${lines === 1 ? "" : "s"}]`;
@@ -4136,6 +4283,19 @@ function useKeyboard(opts) {
     }
     if (state === "ready") {
       if (busyRef.current) return;
+      if (key.ctrl && char === "v") {
+        const b64 = readClipboardImage();
+        if (!b64) {
+          setNotice("no image in clipboard");
+          return;
+        }
+        const chip = `[Image #${++imageCounter} \xB7 clipboard]`;
+        imageStore.set(chip, b64);
+        historyIndex = -1;
+        setInput((s) => s.slice(0, caret) + chip + s.slice(caret));
+        setCaret((i) => i + chip.length);
+        return;
+      }
       const paletteOpen = input.startsWith("/");
       const matches2 = paletteOpen ? filteredCommands(input) : [];
       const mention = !paletteOpen ? parseMention(input) : null;
@@ -4248,8 +4408,16 @@ function useKeyboard(opts) {
           }
         } else if (trimmed) {
           setNotice(null);
-          const message = expandPastes(trimmed);
-          sendMessage(message);
+          const images = [];
+          let textPart = trimmed;
+          for (const [chip, b64] of imageStore) {
+            if (textPart.includes(chip)) {
+              images.push(b64);
+              textPart = textPart.split(chip).join("").trim();
+            }
+          }
+          const message = expandPastes(textPart) || "Describe the attached image.";
+          sendMessage(message, images.length ? images : void 0);
         }
         clearPasteStore();
         setInput(() => "");
@@ -4283,8 +4451,14 @@ function useKeyboard(opts) {
         for (const chip of pasteStore.keys()) {
           if (before.endsWith(chip) && chip.length > match.length) match = chip;
         }
+        for (const chip of imageStore.keys()) {
+          if (before.endsWith(chip) && chip.length > match.length) match = chip;
+        }
         const cut = match ? match.length : 1;
-        if (match) pasteStore.delete(match);
+        if (match) {
+          pasteStore.delete(match);
+          imageStore.delete(match);
+        }
         setInput((s) => s.slice(0, caret - cut) + s.slice(caret));
         setCaret((i) => Math.max(0, i - cut));
       } else if (char && !key.ctrl && !key.meta && !key.tab) {
@@ -4340,7 +4514,7 @@ async function checkForUpdate() {
 import { Fragment as Fragment3, jsx as jsx13, jsxs as jsxs13 } from "react/jsx-runtime";
 function App() {
   const { exit } = useApp();
-  const cwd = process.cwd().replace(homedir6(), "~").split(sep2).join("/");
+  const cwd = process.cwd().replace(homedir7(), "~").split(sep2).join("/");
   const [cfg, setCfg] = useState5(loadConfig());
   const [models, setModels] = useState5([]);
   const [contexts, setContexts] = useState5(() => cfg.modelContexts ?? {});

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "miii-agent",
-  "version": "0.1.25",
+  "version": "0.1.26",
   "description": "Cursor / Claude Code, but local. An offline AI pair-programmer in your terminal, powered by Ollama. Private by default, free forever.",
   "type": "module",
   "bin": {