feat: Context Memory — SQLite FTS5 workspace intelligence with {{Memory:}} tag

- New js/context-memory.js: SQLite FTS5 search engine with heading-aware chunking
- Three storage modes: browser-only, disk workspace, external memory (IndexedDB)
- {{Memory: Name: id}} tag for defining external context sources
- Use: field in AI/Think/Agent tags for multi-source context retrieval
- Memory card with amber accent: Folder/Files/Rebuild buttons + stats
- Context injection into buildPrompt, generateAndReview, generateAgentFlow
- 📚 Memory button in AI Tags overflow dropdown
- Reuses existing sql.js WASM (zero bundle size increase)
- DOMPurify: allow data-memory-name, data-step attributes
- MEMORY_DB key in storage-keys.js
- 30 Playwright tests (context-memory.spec.js)
- Updated README, Feature Showcase template, changelog

Author: ijbo

README.md

@@ -34,14 +34,14 @@
 | **Desktop** | Native app via Neutralino.js with system tray and offline support |
 | **Code Execution** | 6 languages in-browser: Bash ([just-bash](https://justbash.dev/)), Math (Nerdamer), Python ([Pyodide](https://pyodide.org/)), HTML (sandboxed iframe, `html-autorun` for widgets/quizzes), JavaScript (sandboxed iframe), SQL ([sql.js](https://sql.js.org/) SQLite) · 25+ compiled languages via [Judge0 CE](https://ce.judge0.com): C, C++, Rust, Go, Java, TypeScript, Kotlin, Scala, Ruby, Swift, Haskell, Dart, C#, and more |
 | **Security** | Content Security Policy (CSP), SRI integrity hashes, XSS sanitization (DOMPurify), ReDoS protection, Firestore write-token ownership, API keys via HTTP headers, postMessage origin validation, 8-char passphrase minimum, sandboxed code execution |
-| **AI Document Tags** | `{{AI:}}` text generation, `{{Think:}}` deep reasoning, `{{Image:}}` image generation (Gemini Imagen) — per-card model selector, concurrent block operations |
+| **AI Document Tags** | `{{AI:}}` text generation, `{{Think:}}` deep reasoning, `{{Image:}}` image generation (Gemini Imagen), `{{Memory:}}` context attachment (workspace files + external folders via SQLite FTS5, `Use:` field for multi-source retrieval) — per-card model selector, concurrent block operations |
 | **🔌 API Calls** | `{{API:}}` REST API integration — GET/POST/PUT/DELETE methods, custom headers, JSON body, response stored in `$(api_varName)` variables; inline review panel; toolbar GET/POST buttons |
 | **🔗 Agent Flow** | `{{Agent:}}` multi-step pipeline — define Step 1/2/3, chain outputs, per-card model + search provider selector, live step status indicators (⏳/✅/❌), review combined output |
 | **🔍 Web Search** | Toggle web search for AI — DuckDuckGo (free), Brave Search, Serper.dev; search results injected into LLM context; source citations in responses; per-agent-card search provider selector |
 | **🐧 Linux Terminal** | `{{Linux:}}` tag — two modes: (1) Terminal mode opens full Debian Linux ([WebVM](https://webvm.io)) in new window with `Packages:` field; (2) Compile & Run mode (`Language:` + `Script:`) compiles/executes 25+ languages (C++, Rust, Go, Java, Python, TypeScript, Kotlin, Scala…) via [Judge0 CE](https://ce.judge0.com) with inline output, execution time & memory stats |
 | **❓ Help Mode** | Interactive learning mode — click ❓ Help to highlight all buttons, click any button for description + keyboard shortcut + animated demo video; 50% screen demo panel with fullscreen expand; 16 dedicated demo videos mapped to every toolbar button |
 | **Extras** | Auto-save (localStorage + cloud), table of contents, image paste, 103+ templates (11 categories: AI, Agents, Coding, Creative, Documentation, Maths, PPT, Project, Quiz, Tables, Technical), template variable substitution (`$(varName)` with auto-detect), table spreadsheet tools (sort, filter, stats, chart, add row/col, inline cell edit, CSV/MD export), content statistics, modular codebase (13+ JS modules), fully responsive mobile UI with scrollable Quick Action Bar (Files, Search, TOC, Share, Copy, Tools, AI, Model, Upload, Help) and formatting toolbar, multi-file workspace sidebar, disk-backed folder workspace (File System Access API — open folder, save `.md` files to disk, refresh from disk, auto-reconnect), compact header mode with collapsible Tools dropdown (Presentation, Zen, Word Wrap, Focus, Voice, Dark Mode, Preview Theme) |
-| **Dev Tooling** | ESLint + Prettier (lint, format:check), Playwright test suite — 111 tests across smoke, feature, integration, dev, performance, and QA categories (import, export, share, view-mode, editor, email-to-self, secure share, startup timing, export integrity, persistence, module loading, disk workspace, build validation, load-time, accessibility), pre-commit changelog enforcement, GitHub Actions CI |
+| **Dev Tooling** | ESLint + Prettier (lint, format:check), Playwright test suite — 141 tests across smoke, feature, integration, dev, performance, and QA categories (import, export, share, view-mode, editor, email-to-self, secure share, startup timing, export integrity, persistence, module loading, disk workspace, context memory, build validation, load-time, accessibility), pre-commit changelog enforcement, GitHub Actions CI |
 
 ## 🤖 AI Assistant
 
@@ -373,6 +373,7 @@ TextAgent has undergone significant evolution since its inception. What started
 
 | Date | Feature / Update |
 |------|-----------------|
+| **2026-03-10** | 📚 **Context Memory** — new `{{Memory:}}` tag and `Use:` field for workspace intelligence; `js/context-memory.js` module with SQLite FTS5 full-text search (heading-aware chunking, ~1500 chars/chunk); three storage modes: browser-only (IndexedDB blob), disk workspace (`.textagent/memory.db`), external memory (IndexedDB blob); `Use: workspace, my-docs` in AI/Think/Agent tags auto-retrieves relevant context from indexed files; amber-accented Memory card with Folder/Files/Rebuild buttons + stats display; reuses existing sql.js WASM (zero bundle size increase); 30 new Playwright tests (141 total) |
 | **2026-03-10** | 🗂️ **Action Modal & Disk UI Polish** — replaced native `confirm()` and inline rename with unified `showActionModal()` for rename (input field, auto-selects filename), duplicate (blue confirmation), and delete (red destructive); header-only disk controls (refresh ↻, disconnect ✕) replacing footer bar; clickable folder name opens folder picker; same-name rename guard with toast feedback; duplicate tree auto-refresh after disk write; merged CI changelog check into deploy workflow (3→2 workflow runs per push); 10 new Playwright tests (112 total) |
 | **2026-03-10** | 📂 **Disk-Backed Workspace** — new folder storage mode via File System Access API; "Open Folder" button in sidebar header; `.md` files read/written directly to disk; `.textagent/workspace.json` manifest; debounced autosave to disk ("💾 Saved to disk" indicator); refresh from disk for external changes; disconnect to revert to localStorage; auto-reconnect on reload via IndexedDB-stored handles; Chromium-only (hidden in unsupported browsers); 22 new Playwright tests |
 | **2026-03-10** | 🧹 **Clear Text Buttons** — new "Clear" group in formatting toolbar with Clear All (eraser icon) and Clear Selection (backspace icon); custom in-app confirmation modal with smooth animation (replaces flickering native `confirm()`); both undoable via Ctrl+Z; toast feedback; red-accented styling with dark/light theme; Help mode entries |

changelogs/CHANGELOG-context-memory.md

@@ -0,0 +1,34 @@
+# Context Memory — SQLite FTS5 Workspace Intelligence
+
+- New `js/context-memory.js` module: SQLite FTS5-powered search engine for local context
+- Heading-aware chunking (H1/H2/H3 hierarchy, ~1500 char max per chunk)
+- Three storage modes: browser-only (IndexedDB blob), disk workspace (`.textagent/memory.db`), external memory (IndexedDB blob keyed by name)
+- Incremental indexing: tracks file modification times, re-indexes only changed files
+- Reuses existing `sql.js` WASM from `exec-sandbox.js` — no new bundle size increase
+- `{{Memory: Name: my-docs}}` tag for defining external context sources
+- `Use: workspace, my-docs` field in AI/Think/Agent tags for multi-source context retrieval
+- Memory card renders with amber/orange accent: 📂 Folder, 📄 Files, 🔄 Rebuild buttons + stats display
+- Use: hint badge on AI/Think/Agent cards shows active memory sources
+- Context injection: memory search results injected into `buildPrompt()`, `generateAndReview()`, and `generateAgentFlow()` per-step
+- 30 Playwright tests covering module API, tag parsing, card rendering, DOMPurify allowlist
+
+---
+
+## Summary
+Added `{{Memory:}}` tag and `Use:` field for workspace intelligence. AI tags can now leverage indexed local context from workspace files or external folders via SQLite FTS5 full-text search. Three persistence modes ensure memory survives page reloads. Reuses existing sql.js — zero additional bundle size.
+
+---
+
+## Files Changed (9 total)
+
+| File | Lines Changed | Type |
+|------|:---:|------|
+| `js/context-memory.js` | +447 | New module: FTS5 engine, chunking, 3 storage modes |
+| `js/exec-sandbox.js` | +4 | Expose `getSqlJs` on `M._exec` |
+| `js/ai-docgen.js` | +95 −10 | Memory tag parsing, Use: field, Memory card HTML, bindings |
+| `js/ai-docgen-generate.js` | +35 −3 | Memory context injection into prompts |
+| `js/renderer.js` | +1 −1 | DOMPurify: allow `data-memory-name`, `data-step` |
+| `js/storage-keys.js` | +3 | `MEMORY_DB` key |
+| `src/main.js` | +3 | Import `context-memory.js` in Phase 3b-ext |
+| `css/ai-docgen.css` | +90 | Amber Memory card styles, Use: badge |
+| `tests/feature/context-memory.spec.js` | +340 | 30 Playwright tests |

css/ai-docgen.css

@@ -1861,4 +1861,94 @@
 
 [data-theme="light"] .fmt-group-dropdown .fmt-btn:hover {
     background: rgba(0, 0, 0, 0.04);
+}
+
+/* ============================================
+   Memory Card Styles — Amber/Orange accent
+   ============================================ */
+
+/* Memory card — amber accent */
+.ai-placeholder-card[data-ai-type="Memory"] {
+    border-color: rgba(245, 158, 11, 0.4);
+    background: linear-gradient(135deg, rgba(245, 158, 11, 0.08), rgba(217, 119, 6, 0.04));
+}
+
+.ai-placeholder-card[data-ai-type="Memory"] .ai-placeholder-header {
+    border-bottom-color: rgba(245, 158, 11, 0.2);
+}
+
+.ai-placeholder-card[data-ai-type="Memory"] .ai-placeholder-icon {
+    background: rgba(245, 158, 11, 0.15);
+}
+
+.ai-placeholder-card[data-ai-type="Memory"] .ai-placeholder-label {
+    color: #f59e0b;
+}
+
+.ai-placeholder-card[data-ai-type="Memory"] .ai-placeholder-btn {
+    border-color: rgba(245, 158, 11, 0.3);
+    color: #f59e0b;
+}
+
+.ai-placeholder-card[data-ai-type="Memory"] .ai-placeholder-btn:hover {
+    background: rgba(245, 158, 11, 0.15);
+}
+
+/* Memory stats display */
+.ai-memory-stats {
+    padding: 8px 14px;
+    font-size: 0.78em;
+    color: rgba(245, 158, 11, 0.7);
+    border-top: 1px solid rgba(245, 158, 11, 0.1);
+    font-family: 'SF Mono', 'Cascadia Code', monospace;
+}
+
+/* Use: hint badge on AI/Think/Agent cards */
+.ai-use-hint {
+    display: inline-flex;
+    align-items: center;
+    gap: 4px;
+    padding: 2px 8px;
+    border-radius: 12px;
+    font-size: 0.72em;
+    font-weight: 600;
+    background: rgba(245, 158, 11, 0.12);
+    color: #f59e0b;
+    border: 1px solid rgba(245, 158, 11, 0.25);
+    margin-left: 8px;
+    white-space: nowrap;
+}
+
+/* Memory toolbar button */
+.fmt-memory-btn {
+    color: #f59e0b !important;
+}
+
+.fmt-memory-btn:hover {
+    background: rgba(245, 158, 11, 0.15) !important;
+}
+
+/* Light theme overrides */
+[data-theme="light"] .ai-placeholder-card[data-ai-type="Memory"] {
+    border-color: rgba(217, 119, 6, 0.35);
+    background: linear-gradient(135deg, rgba(245, 158, 11, 0.06), rgba(217, 119, 6, 0.03));
+}
+
+[data-theme="light"] .ai-placeholder-card[data-ai-type="Memory"] .ai-placeholder-label {
+    color: #d97706;
+}
+
+[data-theme="light"] .ai-placeholder-card[data-ai-type="Memory"] .ai-placeholder-btn {
+    color: #d97706;
+    border-color: rgba(217, 119, 6, 0.3);
+}
+
+[data-theme="light"] .ai-memory-stats {
+    color: rgba(217, 119, 6, 0.7);
+}
+
+[data-theme="light"] .ai-use-hint {
+    background: rgba(217, 119, 6, 0.1);
+    color: #d97706;
+    border-color: rgba(217, 119, 6, 0.2);
 }

index.html

@@ -622,6 +622,8 @@ <h5>Menu</h5>
                             title="Tag as AI Image Generate">Image</button>
                         <button class="fmt-btn fmt-agent-btn" data-action="agent-tag"
                             title="Tag as Agent Flow Pipeline">Agent</button>
+                        <button class="fmt-btn fmt-memory-btn" data-action="memory-tag"
+                            title="Tag as Context Memory Source">📚 Memory</button>
                         <button class="fmt-btn fmt-fill-btn" data-action="fill-all" id="docgen-fill-btn"
                             title="Fill all AI blocks">Fill</button>
                     </div>

js/ai-docgen-generate.js

@@ -11,14 +11,21 @@
     // PROMPT BUILDING
     // ==============================================
 
-    function buildPrompt(block, prevSections) {
+    function buildPrompt(block, prevSections, memoryContext) {
         var base = block.type === 'Think'
             ? 'You are a thoughtful writer. Think carefully about the topic, then produce polished markdown content.\n\n'
             + 'IMPORTANT: Output ONLY the final polished markdown content. Do NOT include your thinking process, '
             + 'reasoning steps, analysis, or any text like "Thinking Process:" or "Drafting:". '
             + 'Just write the final, high-quality markdown article directly.\n\n'
             : 'You are an expert writer. Write well-formatted, high-quality markdown content for this section.\n\n';
 
+        var retrieval = '';
+        if (memoryContext) {
+            retrieval = '### RELEVANT CONTEXT FROM ATTACHED DOCUMENTS ###\n'
+                + memoryContext + '\n---\n\n'
+                + 'Use the above context to inform your response. Cite file names when relevant.\n\n';
+        }
+
         var instructions = 'Topic/Instructions:\n' + block.prompt + '\n\n';
 
         var context = '';
@@ -27,7 +34,7 @@
                 + prevSections.substring(0, 3000) + '\n\n---\n\n';
         }
 
-        return base + instructions + context
+        return base + retrieval + instructions + context
             + 'Write ONLY the markdown content. Do not include any meta-commentary, explanations, '
             + 'thinking process, or notes about what you wrote. Start directly with the content.';
     }
@@ -439,10 +446,21 @@
                 if (block.type === 'Image') {
                     result = await generateImageForBlock(block, perCardModel);
                 } else {
+                    // Retrieve memory context if Use: field is present
+                    var memoryContext = '';
+                    if (block.useMemory && block.useMemory.length > 0 && M._memory) {
+                        try {
+                            var chunks = await M._memory.search(block.useMemory, block.prompt, 5);
+                            memoryContext = M._memory.formatForContext(chunks);
+                        } catch (memErr) {
+                            console.warn('Memory search failed:', memErr);
+                        }
+                    }
+
                     result = await M.requestAiTask({
                         taskType: 'generate',
                         context: prevContent.substring(Math.max(0, prevContent.length - 3000)),
-                        userPrompt: buildPrompt(block, prevContent),
+                        userPrompt: buildPrompt(block, prevContent, memoryContext),
                         enableThinking: block.type === 'Think',
                         silent: true
                     });
@@ -550,6 +568,18 @@
             }
 
             stepPrompt += 'Task: ' + steps[i].description + '\n\n';
+
+            // Retrieve memory context for this step
+            if (block.useMemory && block.useMemory.length > 0 && M._memory) {
+                try {
+                    var memChunks = await M._memory.search(block.useMemory, steps[i].description, 3);
+                    var memCtx = M._memory.formatForContext(memChunks);
+                    if (memCtx) {
+                        stepPrompt += '### RELEVANT CONTEXT FROM ATTACHED DOCUMENTS ###\n' + memCtx + '\n---\n\n';
+                    }
+                } catch (_) { /* memory search failed, proceed without */ }
+            }
+
             if (accumulatedContext) {
                 stepPrompt += 'Previous steps produced:\n' + accumulatedContext.substring(0, 3000) + '\n\n';
             }