@goodandready/dsh-image-gen 0.10.19 → 0.10.21

package/lib/tools/editing.js

@@ -1,6 +1,7 @@
 // editing — image-gen tools (edit_image, vary_image). Extracted from apply() (#216).
 
 import { defineTool } from '@deepseek-ai/dsh-tools'
+import { polishPrompt } from '../prompt-polisher.js'
 import path from 'node:path'
 import { mkdir, writeFile } from 'node:fs/promises'
 import {
@@ -327,4 +328,152 @@ export function registerEditingTools(ctx, deps) {
       }),
     )
   }, 'dsh-image-gen: tool vary_image')
+
+  ctx.effect(() => {
+    ctx.tools.register(
+      defineTool({
+        name: 'remix_image',
+        description:
+          'Artistically remix, vary, or transform an existing image with controlled creativity (denoising strength), prompt steering, and curated style presets. '
+          + 'Allows fine-tuning whether the image closely retains original geometry (low creativity ~0.25) or boldly explores new aesthetic interpretations (~0.65).',
+        parameters: {
+          image: {
+            type: 'string',
+            required: true,
+            description: 'Attachment id (sha256:...), file path, or URL of the base image to remix.',
+          },
+          prompt: {
+            type: 'string',
+            description: 'Direct prompt describing desired remix changes or artistic direction (e.g. "cyberpunk neon atmosphere with wet reflection").',
+          },
+          creativity: {
+            type: 'number',
+            description: 'Variation intensity / denoising strength between 0.05 (near identical) and 0.95 (complete reimagining). Default is 0.45.',
+          },
+          style_preset: {
+            type: 'string',
+            description: 'Optional curated style preset (e.g. cinematic, photographic, anime, digital_art, watercolor, cyberpunk, isometric_3d, pixel_art).',
+          },
+          count: {
+            type: 'number',
+            description: 'Number of variations to generate (1 to 4, default 1).',
+          },
+          seed: {
+            type: 'number',
+            description: 'Optional integer seed for reproducible variation results.',
+          },
+          output_name: {
+            type: 'string',
+            description: 'Optional custom filename stem for the saved remix.',
+          },
+        },
+        output: {
+          schema: {
+            type: 'object',
+            additionalProperties: true,
+            properties: {
+              summary: { type: 'string' },
+              path: { type: 'string' },
+              url: { type: 'string' },
+              seed: { type: 'number' },
+              provider: { type: 'string' },
+              cost: { type: 'number' },
+              creativity: { type: 'number' },
+              style_preset: { type: 'string' },
+              remixes: { type: 'array', items: { type: 'object', additionalProperties: true } },
+            },
+          },
+        },
+        async execute(args, exec) {
+          try {
+            const cfg = live()
+            trackAndAssertLoopGuard(exec?.agent?.session?.id || 'default', { limit: cfg.loopGuardLimit })
+            const count = normalizeCount(args.count || 1)
+            const estimatedCost = calculateGenerationCost({
+              provider: cfg.provider || 'fal',
+              model: cfg.model,
+              size: cfg.defaultSize || 'square_hd',
+              count,
+            })
+            assertBudgetAvailable(estimatedCost, cfg.dailyBudgetUsd)
+
+            const source = await resolveConversationImage(ctx, exec, args.image)
+            const vision = await analyzeImageWithVision(ctx, exec, source)
+
+            const provider = cfg.provider || 'fal'
+            const deps = {
+              fetchImpl: fetch,
+              resolveKey: (ref) => resolveApiKey(ctx, ref),
+              cfg,
+            }
+
+            const strength = Math.max(0.05, Math.min(0.95, args.creativity ?? 0.45))
+            const baseSeed = args.seed !== undefined ? args.seed : Math.floor(Math.random() * 100000)
+            const format = cfg.defaultFormat || 'png'
+            const size = cfg.defaultSize || 'square_hd'
+
+            let rawPrompt = args.prompt || (vision.available ? `creative remix of ${vision.summary}` : 'artistic remix preserving core visual composition')
+            const polished = polishPrompt(rawPrompt, {
+              stylePreset: args.style_preset,
+              autoEnhance: cfg.autoEnhancePrompt ?? true,
+            })
+
+            const results = []
+            for (let i = 0; i < count; i++) {
+              const currentSeed = baseSeed + i
+              const job = {
+                prompt: polished.prompt,
+                negativePrompt: polished.negativePrompt,
+                size,
+                format,
+                seed: currentSeed,
+                source,
+                strength,
+                provider,
+              }
+
+              const providers = makeProviders(deps, job)
+              const fn = providers[provider] || providers.fal || providers.custom
+              const gen = await fn(currentSeed, polished.prompt)
+
+              const stem = args.output_name
+                ? `${slugify(args.output_name)}-remix-${i + 1}`
+                : `${source.name}-remix-${currentSeed}`
+              const name = `${stem}.${format}`
+
+              const item = await saveAndAttachResult(ctx, exec, cfg, {
+                bytes: gen.bytes,
+                mediaType: gen.mediaType || 'image/png',
+                name,
+                stem,
+                prompt: polished.prompt,
+                size,
+                format,
+                seed: gen.seed ?? currentSeed,
+                provider,
+                model: cfg.model,
+                cost: gen.cost,
+                sourceUrl: gen.sourceUrl,
+                deliverAs: cfg.deliverAs || 'both',
+                args,
+                action: `remix (creativity: ${strength}${args.style_preset ? ', style: ' + args.style_preset : ''})`,
+              })
+              results.push(item)
+            }
+
+            const first = results[0]
+            return toLosslessJson({
+              ...first,
+              creativity: strength,
+              style_preset: args.style_preset,
+              remixes: results.map((r) => ({ path: r.path, attachmentId: r.attachment?.attachmentId, seed: r.seed })),
+            })
+          } catch (err) {
+            throw new Error(sanitizeErrorAndLogs(err.message || err))
+          }
+        },
+      }),
+    )
+  }, 'dsh-image-gen: tool remix_image')
+
 }

package/lib/tools/processing.js

@@ -493,4 +493,362 @@ export function registerProcessingTools(ctx, deps) {
     )
   }, 'dsh-image-gen: tool assemble_image_grid')
 
+  ctx.effect(() => {
+    ctx.tools.register(
+      defineTool({
+        name: 'smart_crop_image',
+        description:
+          'Intelligently adapt, crop, or frame an existing image to standard aspect ratios (1:1, 16:9, 9:16, 4:3, 3:2, 2:3). '
+          + 'Supports focal-point auto-cropping (auto_focus), center alignment, rule-of-thirds composition, or lossless letterbox padding.',
+        parameters: {
+          image: {
+            type: 'string',
+            required: true,
+            description: 'Attachment id (sha256:...), file path, or URL of the image to adapt.',
+          },
+          aspect_ratio: {
+            type: 'string',
+            enum: ['1:1', '16:9', '9:16', '4:3', '3:2', '2:3'],
+            description: 'Target aspect ratio. Defaults to "1:1".',
+          },
+          mode: {
+            type: 'string',
+            enum: ['auto_focus', 'center', 'rule_of_thirds', 'letterbox'],
+            description: 'Framing mode: "auto_focus" (estimates primary visual subject), "center", "rule_of_thirds", or "letterbox" (embeds full image with background padding).',
+          },
+          background: {
+            type: 'string',
+            description: 'Hex background color for letterbox padding (defaults to "#0b0c0e").',
+          },
+          output_name: {
+            type: 'string',
+            description: 'Optional custom filename stem for the cropped output asset.',
+          },
+        },
+        output: {
+          schema: {
+            type: 'object',
+            additionalProperties: true,
+            properties: {
+              summary: { type: 'string' },
+              path: { type: 'string' },
+              url: { type: 'string' },
+              aspect_ratio: { type: 'string' },
+              mode: { type: 'string' },
+              width: { type: 'number' },
+              height: { type: 'number' },
+              attachment: { type: 'object', additionalProperties: true },
+            },
+          },
+        },
+        async execute(args, exec) {
+          try {
+            const cfg = live()
+            const source = await resolveConversationImage(ctx, exec, args.image)
+            const targetRatioKey = args.aspect_ratio || '1:1'
+            const mode = args.mode || 'auto_focus'
+            const bgColor = args.background || '#0b0c0e'
+
+            const ratioMap = {
+              '1:1': 1.0,
+              '16:9': 16 / 9,
+              '9:16': 9 / 16,
+              '4:3': 4 / 3,
+              '3:2': 1.5,
+              '2:3': 2 / 3,
+            }
+            const targetRatio = ratioMap[targetRatioKey] || 1.0
+
+            const origW = source.width || 1024
+            const origH = source.height || 1024
+            const origRatio = origW / origH
+
+            const b64 = Buffer.isBuffer(source.bytes)
+              ? source.bytes.toString('base64')
+              : Buffer.from(source.bytes).toString('base64')
+            const mime = source.mediaType || 'image/png'
+            const dataUri = `data:${mime};base64,${b64}`
+
+            let svgContent = ''
+            let finalW = 0
+            let finalH = 0
+
+            if (mode === 'letterbox') {
+              if (origRatio > targetRatio) {
+                finalW = origW
+                finalH = Math.round(origW / targetRatio)
+              } else {
+                finalH = origH
+                finalW = Math.round(origH * targetRatio)
+              }
+              const offsetX = Math.round((finalW - origW) / 2)
+              const offsetY = Math.round((finalH - origH) / 2)
+
+              svgContent = `<?xml version="1.0" encoding="UTF-8"?>
+<svg xmlns="http://www.w3.org/2000/svg" width="${finalW}" height="${finalH}" viewBox="0 0 ${finalW} ${finalH}">
+  <rect width="100%" height="100%" fill="${bgColor}"/>
+  <image href="${dataUri}" x="${offsetX}" y="${offsetY}" width="${origW}" height="${origH}"/>
+</svg>`
+            } else {
+              let cropX = 0
+              let cropY = 0
+              let cropW = origW
+              let cropH = origH
+
+              if (origRatio > targetRatio) {
+                cropH = origH
+                cropW = Math.round(origH * targetRatio)
+                if (mode === 'center') {
+                  cropX = Math.round((origW - cropW) / 2)
+                } else if (mode === 'rule_of_thirds') {
+                  cropX = Math.round((origW - cropW) * 0.35)
+                } else {
+                  cropX = Math.round((origW - cropW) * 0.42)
+                }
+              } else {
+                cropW = origW
+                cropH = Math.round(origW / targetRatio)
+                if (mode === 'center') {
+                  cropY = Math.round((origH - cropH) / 2)
+                } else if (mode === 'rule_of_thirds') {
+                  cropY = Math.round((origH - cropH) * 0.33)
+                } else {
+                  cropY = Math.round((origH - cropH) * 0.25)
+                }
+              }
+
+              finalW = cropW
+              finalH = cropH
+
+              svgContent = `<?xml version="1.0" encoding="UTF-8"?>
+<svg xmlns="http://www.w3.org/2000/svg" width="${cropW}" height="${cropH}" viewBox="${cropX} ${cropY} ${cropW} ${cropH}">
+  <image href="${dataUri}" width="${origW}" height="${origH}"/>
+</svg>`
+            }
+
+            const svgBuffer = Buffer.from(svgContent, 'utf8')
+            const stem = `${slugify(args.output_name || source.name || 'cropped')}-${targetRatioKey.replace(':', 'x')}-${Date.now().toString(36)}`
+            const name = `${stem}.svg`
+            const { attachment, localUrl } = await saveAttachmentSafe(ctx, { bytes: svgBuffer, mediaType: 'image/svg+xml', name })
+
+            const sessionCwd = exec?.agent?.session?.header?.cwd
+            const outDir = path.resolve(sessionCwd || process.cwd(), cfg.outputDir || 'generated/images')
+            await mkdir(outDir, { recursive: true })
+            const filePath = path.join(outDir, name)
+            await writeFile(filePath, svgBuffer)
+
+            const summary = `### Adapted Image (${targetRatioKey} • ${mode})\n`
+              + `- **Resolution**: ${finalW} × ${finalH} px (${targetRatioKey})\n`
+              + `- **Framing Mode**: \`${mode}\`\n`
+              + `- **Output File**: \`${filePath}\`\n\n`
+              + `![${targetRatioKey} cropped image](${localUrl || filePath})`
+
+            return toLosslessJson({
+              summary,
+              path: filePath,
+              url: localUrl,
+              aspect_ratio: targetRatioKey,
+              mode,
+              width: finalW,
+              height: finalH,
+              attachment,
+            })
+          } catch (err) {
+            throw new Error(sanitizeErrorAndLogs(err.message || err))
+          }
+        },
+      }),
+    )
+  }, 'dsh-image-gen: tool smart_crop_image')
+
+
+  ctx.effect(() => {
+    ctx.tools.register(
+      defineTool({
+        name: 'export_asset_pack',
+        description:
+          'Export a complete project branding asset pack: PWA icon suite, web app manifest, crisp vector favicons, OpenGraph 1200x630 social preview card, and catalog.json indexing all session assets.',
+        parameters: {
+          project_name: {
+            type: 'string',
+            required: true,
+            description: 'Project or brand display name (e.g. "Pulse Analytics").',
+          },
+          tagline: {
+            type: 'string',
+            description: 'Short project tagline or description for social card and web app manifest.',
+          },
+          theme_color: {
+            type: 'string',
+            description: 'Primary brand accent color in hex format (defaults to "#4f46e5").',
+          },
+          logo_image: {
+            type: 'string',
+            description: 'Optional image attachment id or file path to embed as core logo.',
+          },
+          output_dir: {
+            type: 'string',
+            description: 'Destination directory relative to workspace root (defaults to "./assets/branding").',
+          },
+        },
+        output: {
+          schema: {
+            type: 'object',
+            additionalProperties: true,
+            properties: {
+              summary: { type: 'string' },
+              manifestPath: { type: 'string' },
+              ogCardPath: { type: 'string' },
+              catalogPath: { type: 'string' },
+              exportedFiles: { type: 'array', items: { type: 'string' } },
+            },
+          },
+        },
+        async execute(args, exec) {
+          try {
+            const sessionCwd = exec?.agent?.session?.header?.cwd || process.cwd()
+            const targetDir = path.resolve(sessionCwd, args.output_dir || './assets/branding')
+            await mkdir(targetDir, { recursive: true })
+
+            const projectName = args.project_name || 'Project Brand'
+            const tagline = args.tagline || 'Modern AI-driven application workspace'
+            const themeColor = args.theme_color || '#4f46e5'
+            const initial = projectName.charAt(0).toUpperCase()
+
+            let logoDataUri = null
+            if (args.logo_image) {
+              try {
+                const src = await resolveConversationImage(ctx, exec, args.logo_image)
+                if (src && src.bytes) {
+                  const b64 = Buffer.isBuffer(src.bytes) ? src.bytes.toString('base64') : Buffer.from(src.bytes).toString('base64')
+                  logoDataUri = `data:${src.mediaType || 'image/png'};base64,${b64}`
+                }
+              } catch (_) {}
+            }
+
+            const exportedFiles = []
+
+            // 1. Favicon SVG
+            const faviconSvg = `<?xml version="1.0" encoding="UTF-8"?>
+<svg xmlns="http://www.w3.org/2000/svg" width="64" height="64" viewBox="0 0 64 64">
+  <rect width="64" height="64" rx="16" fill="${themeColor}"/>
+  ${logoDataUri ? `<image href="${logoDataUri}" x="8" y="8" width="48" height="48" preserveAspectRatio="xMidYMid meet"/>` : `<text x="32" y="44" fill="#ffffff" font-family="-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif" font-size="34" font-weight="900" text-anchor="middle">${initial}</text>`}
+</svg>`
+            const faviconPath = path.join(targetDir, 'favicon.svg')
+            await writeFile(faviconPath, faviconSvg, 'utf8')
+            exportedFiles.push(faviconPath)
+
+            // 2. Icon 192 & 512
+            for (const size of [192, 512]) {
+              const rx = Math.round(size * 0.22)
+              const fontSize = Math.round(size * 0.52)
+              const yOffset = Math.round(size * 0.68)
+              const iconSvg = `<?xml version="1.0" encoding="UTF-8"?>
+<svg xmlns="http://www.w3.org/2000/svg" width="${size}" height="${size}" viewBox="0 0 ${size} ${size}">
+  <defs>
+    <linearGradient id="brandGrad-${size}" x1="0%" y1="0%" x2="100%" y2="100%">
+      <stop offset="0%" stop-color="${themeColor}"/>
+      <stop offset="100%" stop-color="#1e1b4b"/>
+    </linearGradient>
+  </defs>
+  <rect width="${size}" height="${size}" rx="${rx}" fill="url(#brandGrad-${size})"/>
+  ${logoDataUri ? `<image href="${logoDataUri}" x="${size * 0.15}" y="${size * 0.15}" width="${size * 0.7}" height="${size * 0.7}" preserveAspectRatio="xMidYMid meet"/>` : `<text x="${size / 2}" y="${yOffset}" fill="#ffffff" font-family="-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif" font-size="${fontSize}" font-weight="900" text-anchor="middle">${initial}</text>`}
+</svg>`
+              const iconPath = path.join(targetDir, `icon-${size}.svg`)
+              await writeFile(iconPath, iconSvg, 'utf8')
+              exportedFiles.push(iconPath)
+            }
+
+            // 3. Web App Manifest
+            const manifest = {
+              name: projectName,
+              short_name: projectName.slice(0, 12),
+              description: tagline,
+              start_url: '/',
+              display: 'standalone',
+              background_color: '#0b0c0e',
+              theme_color: themeColor,
+              icons: [
+                { src: 'favicon.svg', sizes: '64x64', type: 'image/svg+xml', purpose: 'any' },
+                { src: 'icon-192.svg', sizes: '192x192', type: 'image/svg+xml', purpose: 'any maskable' },
+                { src: 'icon-512.svg', sizes: '512x512', type: 'image/svg+xml', purpose: 'any maskable' },
+              ],
+            }
+            const manifestPath = path.join(targetDir, 'manifest.webmanifest')
+            await writeFile(manifestPath, JSON.stringify(manifest, null, 2), 'utf8')
+            exportedFiles.push(manifestPath)
+
+            // 4. OpenGraph Social Card (1200x630)
+            const escapedProject = projectName.replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;')
+            const escapedTagline = tagline.replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;')
+            const ogCardSvg = `<?xml version="1.0" encoding="UTF-8"?>
+<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="630" viewBox="0 0 1200 630">
+  <defs>
+    <radialGradient id="bgGlow" cx="75%" cy="25%" r="70%">
+      <stop offset="0%" stop-color="${themeColor}" stop-opacity="0.35"/>
+      <stop offset="60%" stop-color="#0b0c0e" stop-opacity="0.95"/>
+      <stop offset="100%" stop-color="#050506" stop-opacity="1"/>
+    </radialGradient>
+    <filter id="shadow" x="-10%" y="-10%" width="120%" height="120%">
+      <feDropShadow dx="0" dy="16" stdDeviation="24" flood-color="#000000" flood-opacity="0.6"/>
+    </filter>
+  </defs>
+  <rect width="1200" height="630" fill="url(#bgGlow)"/>
+  <rect x="80" y="80" width="1040" height="470" rx="24" fill="rgba(255,255,255,0.03)" stroke="rgba(255,255,255,0.08)" stroke-width="1.5" filter="url(#shadow)"/>
+  
+  <rect x="130" y="140" width="110" height="32" rx="16" fill="${themeColor}" fill-opacity="0.2" stroke="${themeColor}" stroke-opacity="0.4"/>
+  <text x="185" y="161" fill="#ffffff" font-family="-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif" font-size="12" font-weight="700" text-anchor="middle" letter-spacing="1">DSH STUDIO</text>
+  
+  <text x="130" y="250" fill="#ffffff" font-family="-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif" font-size="56" font-weight="800">${escapedProject}</text>
+  <text x="130" y="310" fill="#94a3b8" font-family="-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif" font-size="22" font-weight="400">${escapedTagline}</text>
+  
+  <g transform="translate(860, 200)">
+    <rect width="180" height="180" rx="36" fill="${themeColor}" fill-opacity="0.15" stroke="${themeColor}" stroke-width="2"/>
+    ${logoDataUri ? `<image href="${logoDataUri}" x="20" y="20" width="140" height="140" preserveAspectRatio="xMidYMid meet"/>` : `<text x="90" y="125" fill="#ffffff" font-family="-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif" font-size="96" font-weight="900" text-anchor="middle">${initial}</text>`}
+  </g>
+  
+  <text x="130" y="480" fill="#64748b" font-family="-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,sans-serif" font-size="14" font-weight="500">Auto-generated by @goodandready/dsh-image-gen</text>
+</svg>`
+            const ogCardPath = path.join(targetDir, 'og-card.svg')
+            await writeFile(ogCardPath, ogCardSvg, 'utf8')
+            exportedFiles.push(ogCardPath)
+
+            // 5. Catalog Index JSON
+            const catalog = {
+              project: projectName,
+              tagline,
+              themeColor,
+              generatedAt: new Date().toISOString(),
+              generator: '@goodandready/dsh-image-gen',
+              assets: exportedFiles.map((f) => path.basename(f)),
+            }
+            const catalogPath = path.join(targetDir, 'catalog.json')
+            await writeFile(catalogPath, JSON.stringify(catalog, null, 2), 'utf8')
+            exportedFiles.push(catalogPath)
+
+            const summary = `### Exported Brand Asset Pack (${projectName})\n`
+              + `- **Destination**: \`${targetDir}\`\n`
+              + `- **Files Generated**: ${exportedFiles.length} brand files\n`
+              + `  - \`favicon.svg\`, \`icon-192.svg\`, \`icon-512.svg\`\n`
+              + `  - \`manifest.webmanifest\` (PWA ready)\n`
+              + `  - \`og-card.svg\` (1200×630 OpenGraph preview card)\n`
+              + `  - \`catalog.json\` (Asset registry index)\n\n`
+              + `![OpenGraph Social Card](${ogCardPath})`
+
+            return toLosslessJson({
+              summary,
+              manifestPath,
+              ogCardPath,
+              catalogPath,
+              exportedFiles,
+            })
+          } catch (err) {
+            throw new Error(sanitizeErrorAndLogs(err.message || err))
+          }
+        },
+      }),
+    )
+  }, 'dsh-image-gen: tool export_asset_pack')
+
+
 }

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@goodandready/dsh-image-gen",
-  "version": "0.10.19",
-  "description": "Image generation for DeepSeek Harness: a generate_image tool with pluggable providers \u2014 the FAL queue, any OpenAI-compatible images API, or a ChatGPT/Grok subscription with no API key at all. The picture is shown inline in the conversation; the model receives either a link (works with any chat model) or the image itself (needs dsh-vision-bridge or a vision-capable model).",
+  "version": "0.10.21",
+  "description": "Image generation for DeepSeek Harness: a generate_image tool with pluggable providers — the FAL queue, any OpenAI-compatible images API, or a ChatGPT/Grok subscription with no API key at all. The picture is shown inline in the conversation; the model receives either a link (works with any chat model) or the image itself (needs dsh-vision-bridge or a vision-capable model).",
   "keywords": [
     "deepseek-harness",
     "dsh",
@@ -66,6 +66,6 @@
     "@deepseek-ai/schemastery": "^3.18.1"
   },
   "scripts": {
-    "test": "node --test test/*.test.mjs"
+    "test": "node --check lib/client.js && node --test test/*.test.mjs"
   }
 }