Remove linux support

.github/workflows/release.yml

@@ -2,8 +2,8 @@ name: Release
 
 # Manual release: from the GitHub "Actions" tab pick "Release", click
 # "Run workflow", enter the version tag, and it builds the Windows installer
-# .exe and the Linux tarball + .deb, then publishes a GitHub Release with all
-# artifacts attached and auto-generated notes.
+# .exe, then publishes a GitHub Release with that artifact attached and
+# auto-generated notes.
 on:
   workflow_dispatch:
     inputs:
@@ -68,41 +68,9 @@ jobs:
           path: releases/*.exe
           if-no-files-found: error
 
-  build-linux:
-    name: Build Linux tarball + .deb
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/setup-node@v4
-        with:
-          node-version: 20
-          cache: npm
-
-      - name: Install dependencies
-        run: npm ci
-
-      - name: Set version from input
-        shell: bash
-        env:
-          VERSION: ${{ inputs.version }}
-        run: npm version "${VERSION#v}" --no-git-tag-version --allow-same-version
-
-      - name: Package Linux artifacts (tarball + .deb)
-        shell: bash
-        env:
-          STEPFORGE_PACKAGE_DIR: ${{ github.workspace }}/build/artifacts
-        run: bash scripts/package-linux.sh
-
-      - uses: actions/upload-artifact@v4
-        with:
-          name: linux-artifacts
-          path: build/artifacts/*
-          if-no-files-found: error
-
   release:
     name: Publish GitHub Release
-    needs: [build-windows, build-linux]
+    needs: [build-windows]
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
@@ -127,7 +95,6 @@ jobs:
           fi
           gh release create "$TAG" \
             dist/windows-installer/*.exe \
-            dist/linux-artifacts/* \
             --title "StepForge $TAG" \
             --target "${{ github.sha }}" \
             --generate-notes \

README.md

@@ -1,9 +1,11 @@
 # StepForge
 
-StepForge is a **fully offline**, open-source desktop app for Windows and
-Linux that captures step-by-step workflows as screenshots, lets you annotate
-and describe each step in a focused three-pane editor, and exports the result
-to Markdown, DOCX, PPTX, PDF, HTML (WIP), GIF (WIP), confluence (WIP), Wiki.js (WIP), and image bundles (WIP). The current reconmendations for exporting is Markdown and PDF.
+StepForge is a **fully offline**, open-source desktop app for Windows, with
+Linux (WIP) builds. It captures step-by-step workflows as screenshots, lets
+you annotate and describe each step in a focused three-pane editor, and
+exports the result to Markdown, DOCX, PPTX, PDF, HTML (WIP), GIF (WIP),
+confluence (WIP), Wiki.js (WIP), and image bundles (WIP). The current
+reconmendations for exporting is Markdown and PDF.
 
 It is an independent offline desktop guide-capture tool inspired by publicly
 documented workflow patterns of commercial documentation tools like Folge. It contains no
@@ -60,7 +62,7 @@ using only Node built-ins.
 
 ## Getting Started
 
-For a windows installation, see [docs/windows_installation](docs/windows_installation.md) or for a developer/more in depth walkthrough, see [docs/GETTING_STARTED.md](docs/GETTING_STARTED.md).
+For a Windows installation, see [docs/windows_installation](docs/windows_installation.md) or for a developer/more in depth walkthrough, see [docs/GETTING_STARTED.md](docs/GETTING_STARTED.md).
 
 Requirements: Node.js 20+ and npm (Electron is the only dependency).
 
@@ -70,7 +72,8 @@ npm start          # launch StepForge
 ```
 
 First run creates the local data directory (`~/.local/share/stepforge` on
-Linux, `%APPDATA%/stepforge` on Windows; override with `STEPFORGE_DATA_DIR`).
+Linux (WIP), `%APPDATA%/stepforge` on Windows; override with
+`STEPFORGE_DATA_DIR`).
 
 ## Testing
 
@@ -92,7 +95,7 @@ documents, and validating the bytes of the output, not string matching.
 bash scripts/bootstrap-offline.sh   # verify toolchain availability
 bash scripts/verify.sh              # full test suite + smoke checks
 bash scripts/build-release.sh       # assemble runnable app directory
-bash scripts/package-linux.sh       # portable tar.gz + .deb (+ AppDir spec)
+bash scripts/package-linux.sh       # local Linux packaging (WIP; not part of release)
 npm run package:windows             # Windows installer .exe in releases/
 pwsh scripts/package-windows.ps1    # same Windows installer build via PowerShell
 ```

TODO.md

@@ -1 +1 @@
-In the future, planned updates include a re-vamp of the settings panel, text box's (important, tip, warning, and note) are not properly moving to the intended areas and I will look into locking the size (or being able to control the size) of images throughout varias export formats. OCR or local ai would be pretty cool....
+In the future, planned updates include a re-vamp of the settings panel, text box's (important, tip, warning, and note) are not properly moving to the intended areas and I will look into locking the size (or being able to control the size) of images throughout varias export formats.

app/renderer/dialogs.js

@@ -318,13 +318,8 @@ function showSettingsDialog({
     const aiAutoDoc = el('input', { type: 'checkbox', checked: Boolean(settings.ai?.autoDoc) });
     const ollamaHost = makeInput(settings.ai?.ollama?.host || 'http://127.0.0.1:11434');
     const ollamaModel = makeInput(settings.ai?.ollama?.model || 'llama3.2:1b');
-    const aiStatus = el('div', { className: 'muted ai-status' }, 'AI stays local through Ollama. Vision-capable models can also inspect the screenshot attached to each step.');
+    const aiStatus = el('div', { className: 'muted ai-status' }, 'AI stays local through Ollama. Nothing is sent to the cloud.');
     const testAiBtn = el('button', { type: 'button' }, 'Test connection');
-    const persistOllamaModel = debounce(() => {
-      const model = ollamaModel.value.trim();
-      // Keep the last model choice even if the dialog is dismissed without a full save.
-      void api.settings.set({ keyPath: 'ai.ollama.model', value: model }).catch(() => {});
-    }, 250);
 
     const updateAiStatus = (message, { error = false } = {}) => {
       aiStatus.textContent = message;
@@ -346,9 +341,7 @@ function showSettingsDialog({
           return;
         }
         if (result.installed) {
-          updateAiStatus(result.vision
-            ? `Connected to ${result.host} with ${result.model}. It can inspect screenshots.`
-            : `Connected to ${result.host} with ${result.model}. This model is text-only, so StepForge will use OCR and metadata only.`);
+          updateAiStatus(`Connected to ${result.host} with ${result.model}.`);
         } else {
           updateAiStatus(`Connected to ${result.host}. Model ${result.model} is not installed yet.`, { error: true });
         }
@@ -358,8 +351,6 @@ function showSettingsDialog({
         setButtonLoading(testAiBtn, false);
       }
     };
-    ollamaModel.addEventListener('input', () => persistOllamaModel());
-    ollamaModel.addEventListener('blur', () => persistOllamaModel.flush());
 
     const placeholderRows = el('div', { className: 'placeholder-rows' });
     const rows = [];

app/text-intel.js

@@ -35,15 +35,6 @@ function clamp(v, min, max) {
   return Math.min(max, Math.max(min, v));
 }
 
-function modelLooksVisionCapable(model) {
-  const clean = normalizeWhitespace(model).toLowerCase();
-  if (!clean) return false;
-  return clean.includes('vision')
-    || clean.includes('llava')
-    || clean.includes('gemma4')
-    || /(^|[^a-z0-9])qwen[23](?:\.[0-9]+)?vl([^a-z0-9]|$)/.test(clean);
-}
-
 let createWorkerImpl = null;
 function loadCreateWorker() {
   if (createWorkerImpl) return createWorkerImpl;
@@ -73,7 +64,6 @@ class TextIntelService {
     this.workerPromise = null;
     this.workerQueue = Promise.resolve();
     this.ocrDataDir = path.join(dataDir, 'ocr', 'eng');
-    this.modelCapabilityCache = new Map();
   }
 
   async shutdown() {
@@ -382,63 +372,15 @@ public static class Win32 {
     const data = await res.json();
     const models = Array.isArray(data?.models) ? data.models.map((model) => model.name).filter(Boolean) : [];
     const installed = config.ollama.model ? models.includes(config.ollama.model) : false;
-    const vision = installed ? await this.modelSupportsVision({
-      host: config.ollama.host,
-      model: config.ollama.model,
-    }) : false;
     return {
       ok: true,
       installed,
-      vision,
       models,
       host: config.ollama.host,
       model: config.ollama.model,
     };
   }
 
-  async modelCapabilities({ host, model }) {
-    const normalizedHost = normalizeOllamaHost(host);
-    const normalizedModel = normalizeWhitespace(model);
-    if (!normalizedHost || !normalizedModel) return [];
-    const cacheKey = `${normalizedHost}::${normalizedModel}`;
-    if (this.modelCapabilityCache.has(cacheKey)) {
-      return this.modelCapabilityCache.get(cacheKey);
-    }
-    const url = new URL('/api/show', `${normalizedHost.replace(/\/+$/, '')}/`);
-    let capabilities = [];
-    try {
-      const response = await this.fetch(url, {
-        method: 'POST',
-        headers: { 'Content-Type': 'application/json' },
-        body: JSON.stringify({ model: normalizedModel }),
-      });
-      if (response.ok) {
-        const payload = await response.json();
-        capabilities = Array.isArray(payload?.capabilities)
-          ? payload.capabilities.map((cap) => normalizeWhitespace(cap).toLowerCase()).filter(Boolean)
-          : [];
-      }
-    } catch {
-      capabilities = [];
-    }
-    if (!capabilities.includes('vision') && modelLooksVisionCapable(normalizedModel)) {
-      capabilities = [...capabilities, 'vision'];
-    }
-    this.modelCapabilityCache.set(cacheKey, capabilities);
-    return capabilities;
-  }
-
-  async modelSupportsVision({ host, model }) {
-    const capabilities = await this.modelCapabilities({ host, model });
-    return capabilities.includes('vision');
-  }
-
-  readStepImageBase64(guideId, stepId) {
-    const imagePath = this.store.stepImagePath(guideId, stepId, 'working') || this.store.stepImagePath(guideId, stepId, 'original');
-    if (!imagePath || !fs.existsSync(imagePath)) return '';
-    return fs.readFileSync(imagePath).toString('base64');
-  }
-
   async callOllamaText({ host, model, prompt, systemPrompt }) {
     const url = new URL('/api/chat', `${host.replace(/\/+$/, '')}/`);
     const response = await this.fetch(url, {
@@ -461,12 +403,8 @@ public static class Win32 {
     return content.trim();
   }
 
-  async callOllama({ host, model, prompt, systemPrompt, images = [] }) {
+  async callOllama({ host, model, prompt, systemPrompt }) {
     const url = new URL('/api/chat', `${host.replace(/\/+$/, '')}/`);
-    const userMessage = { role: 'user', content: prompt };
-    if (Array.isArray(images) && images.length) {
-      userMessage.images = images;
-    }
     const response = await this.fetch(url, {
       method: 'POST',
       headers: { 'Content-Type': 'application/json' },
@@ -476,7 +414,7 @@ public static class Win32 {
         format: 'json',
         messages: [
           { role: 'system', content: systemPrompt },
-          userMessage,
+          { role: 'user', content: prompt },
         ],
         options: {
           temperature: 0.2,
@@ -522,14 +460,6 @@ public static class Win32 {
         return { ok: false, reason: 'Block not found.' };
       }
 
-      const screenshotBase64 = step.image ? this.readStepImageBase64(guideId, stepId) : '';
-      const screenshotAttached = Boolean(screenshotBase64)
-        ? await this.modelSupportsVision({
-          host: config.ollama.host,
-          model: config.ollama.model,
-        })
-        : false;
-
       let captureContext = null;
       // Use stored capture metadata when available (best context, from capture time).
       // Fall back to re-running OCR on the stored image only when metadata is absent.
@@ -584,7 +514,6 @@ public static class Win32 {
         step,
         captureContext,
         block: currentBlock,
-        screenshotAttached,
       });
 
       const raw = await this.callOllama({
@@ -592,7 +521,6 @@ public static class Win32 {
         model: config.ollama.model,
         prompt,
         systemPrompt,
-        images: screenshotAttached ? [screenshotBase64] : [],
       });
       const patch = normalizeAiPatch(raw);
       const updated = applyAiPatchToStep(step, patch, { target, blockId });

core/text-intel.js

@@ -673,7 +673,6 @@ function buildAiPrompt({
   step = null,
   captureContext = null,
   block = null,
-  screenshotAttached = false,
 } = {}) {
   const hasDraftTitle = step && !isPlaceholderTitle(step.title);
   const hasDraftDesc = step && Boolean(htmlToText(step.descriptionHtml || ''));
@@ -718,7 +717,6 @@ function buildAiPrompt({
       (!hasDraftTitle || target === 'description') && captureContext.titleCandidate
         ? `Suggested title: ${captureContext.titleCandidate}` : null,
     ] : []),
-    screenshotAttached ? 'Screenshot: attached to this request.' : null,
     draftTitleLine,
     draftDescLine,
   ].filter(Boolean);
@@ -772,9 +770,6 @@ function buildAiPrompt({
     richContext
       ? '- Use the OCR text, window title, app name, and element info to make the documentation specific.'
       : '- Context is limited. Use the app name or window title if available; generate a reasonable action title.',
-    screenshotAttached
-      ? '- A screenshot is attached. Use it together with the OCR and metadata to resolve visual details, but do not mention the screenshot in the output.'
-      : '- No screenshot is attached. Rely on OCR, the window title, app name, and element info.',
     '- Do NOT generate blocks that describe the technical capture process or mention OCR.',
     '- Do NOT invent details not supported by the capture context.',
     '- If the target is one block, only rewrite that block.',

docs/ARCHITECTURE.md

@@ -111,7 +111,7 @@ click position. Three pieces make that hold:
 
 1. **OS click events** (`app/capture.js`): a low-level mouse hook on Windows
    (`CLICK x y button unixMs` lines), an `xinput test-xi2 --root` watcher on
-   X11. The Linux parser carries event-time `root:` coordinates and merges
+   X11. The Linux (WIP) parser carries event-time `root:` coordinates and merges
    raw/regular twin blocks structurally — there is no time-based debounce
    that could drop fast clicks, only suppression of identical duplicate
    deliveries. Physical coordinates convert to DIP via

docs/GETTING_STARTED.md

@@ -25,9 +25,9 @@ That installs Electron and the local packaging tools used by the scripts.
 npm start
 ```
 
-The first launch creates the local StepForge data directory. On Linux it is
-usually under `~/.local/share/stepforge`. On Windows it is usually under
-`%APPDATA%/stepforge`.
+The first launch creates the local StepForge data directory. On Linux (WIP)
+it is usually under `~/.local/share/stepforge`. On Windows it is usually
+under `%APPDATA%/stepforge`.
 
 ## 3. Create your first guide
 
@@ -95,4 +95,5 @@ If you want to find commands quickly, press `Ctrl+/` for Quick Actions.
 1. `bash scripts/build-release.sh` assembles the offline release layout.
 2. `npm run package:windows` creates the Windows installer `.exe` in
    `releases/`.
-3. `bash scripts/package-linux.sh` creates Linux release artifacts.
+3. `bash scripts/package-linux.sh` creates Linux release artifacts (WIP;
+   local only).

docs/getting_started_with_ai.md

@@ -22,14 +22,6 @@ ollama pull llama3.2:1b
 
 That model is small enough to feel responsive on modest hardware, but still good enough for human-sounding titles and short text blocks.
 
-If you want StepForge to send the screenshot itself to the model, pull a vision-capable model instead:
-
-```bash
-ollama pull gemma3
-```
-
-That model can inspect pictures as well as text, so it is better when you want the AI to read the UI directly from the screenshot.
-
 If you need something even smaller, try:
 
 ```bash
@@ -52,7 +44,7 @@ Set:
 
 * `Enable AI text filling` to on
 * `Ollama host` to your local Ollama server
-* `Ollama model` to `llama3.2:1b` for text-only mode, or `llama3.2-vision` if you want screenshot-aware AI
+* `Ollama model` to `llama3.2:1b` or the smaller model you pulled
 
 The default host is:
 
@@ -81,4 +73,3 @@ You can also use `More -> Generate all text fields with AI` to fill the whole st
 * Capture titles are still generated automatically without AI.
 * AI generation only works when `Enable AI text filling` is turned on.
 * The app always uses local OCR around the click area first, then local AI only when you ask for it.
-* When the selected Ollama model supports vision, StepForge also sends the screenshot to the model so it can cross-check OCR and visual context.

tests/unit/placeholders.test.js

@@ -56,14 +56,12 @@ test('settings persist, deep-merge with defaults, and store global placeholders'
   assert.equal(s1.get('appearance'), DEFAULT_SETTINGS.appearance);
   s1.set('appearance', 'dark');
   s1.set('capture.delayMs', 1500);
-  s1.set('ai.ollama.model', 'qwen3:0.6b');
   s1.setGlobalPlaceholders({ Company: 'Acme', Author: 'Tyler' });
 
   // A fresh instance reads back the changed values merged over defaults.
   const s2 = new Settings(dir);
   assert.equal(s2.get('appearance'), 'dark');
   assert.equal(s2.get('capture.delayMs'), 1500);
-  assert.equal(s2.get('ai.ollama.model'), 'qwen3:0.6b');
   assert.equal(s2.get('capture.clickMarker'), DEFAULT_SETTINGS.capture.clickMarker);
   assert.deepEqual(s2.getGlobalPlaceholders(), { Company: 'Acme', Author: 'Tyler' });
 });

tests/unit/text-intel.test.js

@@ -1,7 +1,5 @@
 'use strict';
 
-const fs = require('node:fs');
-const path = require('node:path');
 const test = require('node:test');
 const assert = require('node:assert/strict');
 
@@ -311,112 +309,21 @@ test('ollama connection test reports installed models', async (t) => {
     settings: makeSettings(),
     getWindow: () => null,
     dataDir: root,
-    fetchImpl: async (url) => {
-      const pathname = new URL(url).pathname;
-      if (pathname === '/api/tags') {
-        return {
-          ok: true,
-          json: async () => ({
-            models: [
-              { name: 'llama3.2:1b' },
-              { name: 'qwen3:0.6b' },
-            ],
-          }),
-        };
-      }
-      if (pathname === '/api/show') {
-        return {
-          ok: true,
-          json: async () => ({
-            capabilities: ['completion', 'vision'],
-          }),
-        };
-      }
-      throw new Error(`unexpected fetch: ${pathname}`);
-    },
+    fetchImpl: async () => ({
+      ok: true,
+      json: async () => ({
+        models: [
+          { name: 'llama3.2:1b' },
+          { name: 'qwen3:0.6b' },
+        ],
+      }),
+    }),
   });
 
   const result = await service.testAiConnection();
   assert.equal(result.ok, true);
   assert.equal(result.installed, true);
   assert.equal(result.model, 'llama3.2:1b');
-  assert.equal(result.vision, true);
-});
-
-test('vision-capable models receive the screenshot in the chat request', async (t) => {
-  const root = makeTmpDir('text-intel-ai-vision');
-  t.after(() => rmrf(root));
-  const imagePath = path.join(root, 'step.png');
-  fs.writeFileSync(imagePath, Buffer.from('fake screenshot bytes'));
-
-  const step = createStep({
-    title: 'Old title',
-    descriptionHtml: '<p>Old text</p>',
-    image: {
-      originalPath: 'original.png',
-      workingPath: 'working.png',
-      size: { width: 10, height: 10 },
-    },
-    captureMetadata: {
-      windowTitle: 'Settings',
-      appName: 'chrome',
-      ocrText: 'Open settings',
-      titleCandidate: 'Open settings',
-      mode: 'fullscreen',
-    },
-  });
-  const fetchCalls = [];
-  const service = new TextIntelService({
-    store: {
-      settingsDir: root,
-      getGuide: () => ({ guideId: 'g1', title: 'Guide', descriptionHtml: '', stepsOrder: ['s1'] }),
-      getStep: () => step,
-      stepImagePath: () => imagePath,
-      saveStep: (_, next) => next,
-    },
-    settings: makeSettings(),
-    getWindow: () => null,
-    dataDir: root,
-    fetchImpl: async (url, init = {}) => {
-      const pathname = new URL(url).pathname;
-      fetchCalls.push({ pathname, init });
-      if (pathname === '/api/show') {
-        return {
-          ok: true,
-          json: async () => ({
-            capabilities: ['completion', 'vision'],
-          }),
-        };
-      }
-      if (pathname === '/api/chat') {
-        return {
-          ok: true,
-          json: async () => ({
-            message: {
-              content: JSON.stringify({
-                title: 'Open settings',
-                description: 'Use the AI tab.',
-              }),
-            },
-          }),
-        };
-      }
-      throw new Error(`unexpected fetch: ${pathname}`);
-    },
-  });
-
-  const result = await service.generateStepPatch({
-    guideId: 'g1',
-    stepId: 's1',
-    target: 'all',
-  });
-
-  assert.equal(result.ok, true);
-  const chatCall = fetchCalls.find((call) => call.pathname === '/api/chat');
-  assert.ok(chatCall, 'expected an Ollama chat request');
-  const body = JSON.parse(chatCall.init.body);
-  assert.deepEqual(body.messages[1].images, [fs.readFileSync(imagePath).toString('base64')]);
-  assert.match(body.messages[1].content, /Screenshot: attached/i);
 });
 
 test('invalid ollama output fails safely without saving the step', async (t) => {