discoclaw 0.8.0 → 0.8.2

package/dist/cron/executor.js

@@ -1,3 +1,4 @@
+import { resolveDefaultModel as resolveImagegenDefaultModel } from '../discord/actions-imagegen.js';
 import { acquireCronLock, releaseCronLock } from './job-lock.js';
 import { resolveChannel } from '../discord/action-utils.js';
 import { DiscordTransportClient } from '../discord/transport-client.js';
@@ -211,7 +212,7 @@ export async function executeCronJob(job, ctx) {
             });
         // Inject tiered action schema documentation when discord actions are enabled.
         if (ctx.discordActionsEnabled) {
-            const actionSelection = buildTieredDiscordActionsPromptSection(cronActionFlags, ctx.botDisplayName, { userText: job.def.prompt });
+            const actionSelection = buildTieredDiscordActionsPromptSection(cronActionFlags, ctx.botDisplayName, { userText: job.def.prompt, imagegenDefaultModel: ctx.imagegenCtx ? resolveImagegenDefaultModel(ctx.imagegenCtx) : undefined });
             if (actionSelection.prompt) {
                 prompt += '\n\n---\n' + actionSelection.prompt;
             }

package/dist/discord/actions-config.js

@@ -21,6 +21,7 @@ const ROLE_DESCRIPTIONS = {
     cron: 'Cron auto-tagging and model classification',
     'cron-exec': 'Default model for cron job execution (overridden by per-job settings)',
     voice: 'Voice channel AI responses',
+    imagegen: 'Default model for image generation',
 };
 // ---------------------------------------------------------------------------
 // Executor
@@ -229,6 +230,15 @@ export function executeConfigAction(action, configCtx) {
                         return { ok: false, error: 'Voice subsystem not configured' };
                     }
                     break;
+                case 'imagegen':
+                    if (bp.imagegenCtx) {
+                        bp.imagegenCtx.defaultModel = model;
+                        changes.push(`imagegen → ${model}`);
+                    }
+                    else {
+                        return { ok: false, error: 'Imagegen subsystem not configured' };
+                    }
+                    break;
                 default:
                     return { ok: false, error: `Unknown role: ${String(action.role)}` };
             }
@@ -338,6 +348,21 @@ export function executeConfigAction(action, configCtx) {
                             }
                         }
                         break;
+                    case 'imagegen':
+                        if (bp.imagegenCtx) {
+                            // Restore the fallback-resolved default (env setting or provider-based fallback).
+                            const igEnvDefault = defaults['imagegen'];
+                            if (igEnvDefault) {
+                                bp.imagegenCtx.defaultModel = igEnvDefault;
+                            }
+                            else {
+                                // Clear override so resolveDefaultModel falls back to provider detection.
+                                bp.imagegenCtx.defaultModel = undefined;
+                            }
+                            const resolvedIg = resolveDefaultModel(bp.imagegenCtx);
+                            resetChanges.push(`imagegen → ${resolvedIg}`);
+                        }
+                        break;
                 }
                 // Clear the override marker regardless of whether we had a default.
                 if (configCtx.overrideSources) {
@@ -382,7 +407,7 @@ export function executeConfigAction(action, configCtx) {
             if (bp.imagegenCtx) {
                 const igModel = resolveDefaultModel(bp.imagegenCtx);
                 const igProvider = resolveProvider(igModel);
-                rows.push(['imagegen', igModel, `Image generation (${igProvider})`, '']);
+                rows.push(['imagegen', igModel, `Image generation (${igProvider})`, ovr('imagegen')]);
             }
             else {
                 rows.push(['imagegen', 'setup-required', 'Image generation (setup required)', '']);
@@ -474,7 +499,7 @@ export function configActionsPromptSection() {
 <discord-action>{"type":"modelSet","role":"chat","model":"sonnet"}</discord-action>
 <discord-action>{"type":"modelSet","role":"fast","model":"haiku"}</discord-action>
 \`\`\`
-- \`role\` (required): One of \`chat\`, \`plan-run\`, \`fast\`, \`forge-drafter\`, \`forge-auditor\`, \`summary\`, \`cron\`, \`cron-exec\`, \`voice\`.
+- \`role\` (required): One of \`chat\`, \`plan-run\`, \`fast\`, \`forge-drafter\`, \`forge-auditor\`, \`summary\`, \`cron\`, \`cron-exec\`, \`voice\`, \`imagegen\`.
 - \`model\` (required): Model tier (\`fast\`, \`capable\`, \`deep\`), concrete model name (\`haiku\`, \`sonnet\`, \`opus\`), runtime name (\`openrouter\`, \`gemini\` — for \`chat\` and \`voice\` roles, swaps the active runtime adapter independently), or \`default\` (for cron-exec only, to revert to the startup default for that role). For the \`voice\` role, setting a model name that belongs to a different provider's tier map (e.g. \`sonnet\` while voice is on Gemini) will auto-switch the voice runtime to match.
 
 **Roles:**
@@ -489,6 +514,7 @@ export function configActionsPromptSection() {
 | \`cron\` | Cron auto-tagging and model classification (overrides fast) |
 | \`cron-exec\` | Default model for cron job execution; per-job overrides (via \`cronUpdate\`) take priority |
 | \`voice\` | Voice channel AI responses |
+| \`imagegen\` | Default model for image generation |
 
 Changes are **persisted** to \`models.json\` and survive restart. Use \`!models reset\` to clear overrides and revert to defaults.
 

package/dist/discord/actions-config.test.js

@@ -243,7 +243,7 @@ describe('modelShow imagegen row', () => {
         if (!result.ok)
             return;
         expect(result.summary).toContain('imagegen');
-        expect(result.summary).toContain('imagen-4.0-generate-001');
+        expect(result.summary).toContain('gemini-3.1-flash-image-preview');
         expect(result.summary).toContain('gemini');
     });
     it('respects explicit defaultModel', () => {
@@ -267,7 +267,7 @@ describe('modelShow imagegen row', () => {
         expect(result.summary).toContain('setup-required');
         expect(result.summary).toContain('Image generation (setup required)');
     });
-    it('defaults to dall-e-3/openai when both apiKey and geminiApiKey are set', () => {
+    it('defaults to native Gemini when both apiKey and geminiApiKey are set', () => {
         const imagegenCtx = { apiKey: 'sk-test', geminiApiKey: 'gk-test' };
         const ctx = makeCtx({ imagegenCtx });
         const result = executeConfigAction({ type: 'modelShow' }, ctx);
@@ -275,8 +275,8 @@ describe('modelShow imagegen row', () => {
         if (!result.ok)
             return;
         expect(result.summary).toContain('imagegen');
-        expect(result.summary).toContain('dall-e-3');
-        expect(result.summary).toContain('openai');
+        expect(result.summary).toContain('gemini-3.1-flash-image-preview');
+        expect(result.summary).toContain('gemini');
     });
 });
 // ---------------------------------------------------------------------------

package/dist/discord/actions-imagegen.js

@@ -1,6 +1,7 @@
 import { AttachmentBuilder } from 'discord.js';
 import { resolveChannel, findChannelRaw, describeChannelType } from './action-utils.js';
 import { NO_MENTIONS } from './allowed-mentions.js';
+import { downloadMessageImages, downloadImageUrl } from './image-download.js';
 const IMAGEGEN_TYPE_MAP = {
     generateImage: true,
 };
@@ -15,14 +16,19 @@ const GPT_IMAGE_VALID_SIZES = new Set(['1024x1024', '1024x1792', '1792x1024', 'a
 const GEMINI_VALID_SIZES = new Set(['1:1', '3:4', '4:3', '9:16', '16:9']);
 const VALID_QUALITY = new Set(['standard', 'hd']);
 const DISCORD_MAX_CONTENT = 2000;
+// Progress UX
+export const TYPING_INTERVAL_MS = 8_000;
+export const DOT_CYCLE_INTERVAL_MS = 3_000;
+export const REQUEST_TIMEOUT_MS = 120_000;
+const DOT_STATES = ['On it.', 'On it..', 'On it...'];
 // ---------------------------------------------------------------------------
 // Provider resolution
 // ---------------------------------------------------------------------------
 export function resolveDefaultModel(imagegenCtx) {
     if (imagegenCtx.defaultModel)
         return imagegenCtx.defaultModel;
-    if (imagegenCtx.geminiApiKey && !imagegenCtx.apiKey)
-        return 'imagen-4.0-generate-001';
+    if (imagegenCtx.geminiApiKey)
+        return 'gemini-3.1-flash-image-preview';
     return 'dall-e-3';
 }
 export function resolveProvider(model, explicit) {
@@ -37,7 +43,7 @@ export function resolveProvider(model, explicit) {
 // ---------------------------------------------------------------------------
 // API callers
 // ---------------------------------------------------------------------------
-async function callOpenAI(prompt, model, size, quality, apiKey, baseUrl) {
+async function callOpenAI(prompt, model, size, quality, apiKey, baseUrl, signal) {
     const body = {
         model,
         prompt,
@@ -57,6 +63,7 @@ async function callOpenAI(prompt, model, size, quality, apiKey, baseUrl) {
                 'Content-Type': 'application/json',
             },
             body: JSON.stringify(body),
+            signal,
         });
     }
     catch (err) {
@@ -87,7 +94,7 @@ async function callOpenAI(prompt, model, size, quality, apiKey, baseUrl) {
     }
     return { ok: true, b64: imageItem.b64_json };
 }
-async function callGemini(prompt, model, size, geminiApiKey) {
+async function callGemini(prompt, model, size, geminiApiKey, signal) {
     const url = `https://generativelanguage.googleapis.com/v1beta/models/${model}:predict`;
     const body = {
         instances: [{ prompt }],
@@ -105,6 +112,7 @@ async function callGemini(prompt, model, size, geminiApiKey) {
                 'Content-Type': 'application/json',
             },
             body: JSON.stringify(body),
+            signal,
         });
     }
     catch (err) {
@@ -135,10 +143,15 @@ async function callGemini(prompt, model, size, geminiApiKey) {
     }
     return { ok: true, b64 };
 }
-async function callGeminiNative(prompt, model, geminiApiKey) {
+async function callGeminiNative(prompt, model, geminiApiKey, sourceImage, signal) {
     const url = `https://generativelanguage.googleapis.com/v1beta/models/${model}:generateContent`;
+    const parts = [];
+    if (sourceImage) {
+        parts.push({ inlineData: { mimeType: sourceImage.mediaType, data: sourceImage.base64 } });
+    }
+    parts.push({ text: prompt });
     const body = {
-        contents: [{ parts: [{ text: prompt }] }],
+        contents: [{ parts }],
         generationConfig: { responseModalities: ['TEXT', 'IMAGE'] },
     };
     let response;
@@ -150,6 +163,7 @@ async function callGeminiNative(prompt, model, geminiApiKey) {
                 'Content-Type': 'application/json',
             },
             body: JSON.stringify(body),
+            signal,
         });
     }
     catch (err) {
@@ -174,14 +188,57 @@ async function callGeminiNative(prompt, model, geminiApiKey) {
     catch {
         return { ok: false, error: 'generateImage: failed to parse API response' };
     }
-    const parts = data.candidates?.[0]?.content?.parts ?? [];
-    const imagePart = parts.find(p => p.inlineData?.mimeType?.startsWith('image/'));
+    const responseParts = data.candidates?.[0]?.content?.parts ?? [];
+    const imagePart = responseParts.find(p => p.inlineData?.mimeType?.startsWith('image/'));
     if (!imagePart?.inlineData?.data) {
         return { ok: false, error: 'generateImage: API returned no image data' };
     }
     return { ok: true, b64: imagePart.inlineData.data };
 }
 // ---------------------------------------------------------------------------
+// Source image resolution
+// ---------------------------------------------------------------------------
+async function resolveSourceImage(sourceImage, ctx) {
+    if (sourceImage.type === 'url') {
+        const dlResult = await downloadImageUrl(sourceImage.url);
+        if (!dlResult.ok) {
+            return { ok: false, error: `generateImage: ${dlResult.error}` };
+        }
+        return { ok: true, base64: dlResult.image.base64, mediaType: dlResult.image.mediaType };
+    }
+    const channelId = sourceImage.channelId ?? ctx.channelId;
+    const messageId = sourceImage.messageId ?? ctx.messageId;
+    const attachmentIndex = sourceImage.attachmentIndex ?? 0;
+    let channel;
+    try {
+        channel = await ctx.client.channels.fetch(channelId);
+    }
+    catch {
+        return { ok: false, error: `generateImage: could not fetch channel "${channelId}"` };
+    }
+    if (!channel || !('messages' in channel)) {
+        return { ok: false, error: `generateImage: channel "${channelId}" is not a text channel` };
+    }
+    let message;
+    try {
+        message = await channel.messages.fetch(messageId);
+    }
+    catch {
+        return { ok: false, error: `generateImage: could not fetch message "${messageId}"` };
+    }
+    const attachments = [...message.attachments.values()];
+    if (attachmentIndex < 0 || attachmentIndex >= attachments.length) {
+        return { ok: false, error: `generateImage: no attachment at index ${attachmentIndex} (message has ${attachments.length} attachment${attachments.length === 1 ? '' : 's'})` };
+    }
+    const target = attachments[attachmentIndex];
+    const result = await downloadMessageImages([target], 1);
+    if (result.images.length === 0) {
+        const reason = result.errors.length > 0 ? `: ${result.errors[0]}` : '';
+        return { ok: false, error: `generateImage: source image attachment rejected${reason}` };
+    }
+    return { ok: true, base64: result.images[0].base64, mediaType: result.images[0].mediaType };
+}
+// ---------------------------------------------------------------------------
 // Executor
 // ---------------------------------------------------------------------------
 export async function executeImagegenAction(action, ctx, imagegenCtx) {
@@ -238,38 +295,85 @@ export async function executeImagegenAction(action, ctx, imagegenCtx) {
                     return { ok: false, error: 'generateImage: apiKey is required for OpenAI provider' };
                 }
             }
-            // Call provider
-            let result;
-            if (provider === 'gemini') {
-                if (model.startsWith('gemini-')) {
-                    result = await callGeminiNative(action.prompt.trim(), model, imagegenCtx.geminiApiKey);
+            // Source image model gate (sync check, before placeholder)
+            if (action.sourceImage && !model.startsWith('gemini-')) {
+                return { ok: false, error: `generateImage: sourceImage is only supported with native Gemini models (gemini-*), not "${model}"` };
+            }
+            // --- Progress UX lifecycle ---
+            const placeholder = await channel.send({ content: DOT_STATES[0], allowedMentions: NO_MENTIONS });
+            channel.sendTyping().catch(() => { });
+            let dotIndex = 0;
+            const typingInterval = setInterval(() => { channel.sendTyping().catch(() => { }); }, TYPING_INTERVAL_MS);
+            const dotInterval = setInterval(() => {
+                dotIndex = (dotIndex + 1) % DOT_STATES.length;
+                placeholder.edit(DOT_STATES[dotIndex]).catch(() => { });
+            }, DOT_CYCLE_INTERVAL_MS);
+            const controller = new AbortController();
+            const timeoutId = setTimeout(() => controller.abort(), REQUEST_TIMEOUT_MS);
+            try {
+                // Resolve source image if provided
+                let resolvedSourceImage;
+                if (action.sourceImage) {
+                    const srcResult = await resolveSourceImage(action.sourceImage, ctx);
+                    if (!srcResult.ok) {
+                        return { ok: false, error: srcResult.error };
+                    }
+                    resolvedSourceImage = { base64: srcResult.base64, mediaType: srcResult.mediaType };
+                }
+                // Call provider
+                let result;
+                if (provider === 'gemini') {
+                    if (model.startsWith('gemini-')) {
+                        result = await callGeminiNative(action.prompt.trim(), model, imagegenCtx.geminiApiKey, resolvedSourceImage, controller.signal);
+                    }
+                    else {
+                        result = await callGemini(action.prompt.trim(), model, size, imagegenCtx.geminiApiKey, controller.signal);
+                    }
                 }
                 else {
-                    result = await callGemini(action.prompt.trim(), model, size, imagegenCtx.geminiApiKey);
+                    const baseUrl = imagegenCtx.baseUrl ?? 'https://api.openai.com/v1';
+                    result = await callOpenAI(action.prompt.trim(), model, size, quality, imagegenCtx.apiKey, baseUrl, controller.signal);
                 }
+                if (controller.signal.aborted) {
+                    return { ok: false, error: 'generateImage: request timed out' };
+                }
+                if (!result.ok) {
+                    return { ok: false, error: result.error };
+                }
+                // Stop progress before final Discord mutations
+                clearInterval(typingInterval);
+                clearInterval(dotInterval);
+                clearTimeout(timeoutId);
+                await placeholder.delete().catch(() => { });
+                const buf = Buffer.from(result.b64, 'base64');
+                const attachment = new AttachmentBuilder(buf, { name: 'image-1.png' });
+                const sendOpts = { files: [attachment], allowedMentions: NO_MENTIONS };
+                if (action.caption) {
+                    sendOpts.content = action.caption;
+                }
+                await channel.send(sendOpts);
+                return { ok: true, summary: `Generated image posted to #${channel.name}` };
             }
-            else {
-                const baseUrl = imagegenCtx.baseUrl ?? 'https://api.openai.com/v1';
-                result = await callOpenAI(action.prompt.trim(), model, size, quality, imagegenCtx.apiKey, baseUrl);
-            }
-            if (!result.ok) {
-                return { ok: false, error: result.error };
+            catch (err) {
+                const msg = err instanceof Error ? err.message : String(err);
+                return { ok: false, error: `generateImage: ${msg}` };
             }
-            const buf = Buffer.from(result.b64, 'base64');
-            const attachment = new AttachmentBuilder(buf, { name: 'image-1.png' });
-            const sendOpts = { files: [attachment], allowedMentions: NO_MENTIONS };
-            if (action.caption) {
-                sendOpts.content = action.caption;
+            finally {
+                clearInterval(typingInterval);
+                clearInterval(dotInterval);
+                clearTimeout(timeoutId);
+                await placeholder.delete().catch(() => { });
             }
-            await channel.send(sendOpts);
-            return { ok: true, summary: `Generated image posted to #${channel.name}` };
         }
     }
 }
 // ---------------------------------------------------------------------------
 // Prompt section
 // ---------------------------------------------------------------------------
-export function imagegenActionsPromptSection() {
+export function imagegenActionsPromptSection(resolvedDefaultModel) {
+    const modelFieldDoc = resolvedDefaultModel
+        ? `- \`model\` (optional): Default is \`${resolvedDefaultModel}\`. Omit this field to use the default; only set it when a different model is explicitly needed. Supported families/examples:`
+        : `- \`model\` (optional): Default depends on configuration. Supported families/examples:`;
     return `### Image Generation
 
 **generateImage** — Generate an image and post it to a channel:
@@ -278,7 +382,7 @@ export function imagegenActionsPromptSection() {
 \`\`\`
 - \`prompt\` (required): Text description of the image to generate.
 - \`channel\` (optional): Channel name (with or without #) or channel ID to post the image to. Defaults to the current channel/thread if omitted.
-- \`model\` (optional): Model to use. Default depends on configuration (auto-detected from available API keys). Common supported families/examples:
+${modelFieldDoc}
   - OpenAI: \`dall-e-3\`, \`gpt-image-1\`
   - Gemini (Imagen): \`imagen-4.0-generate-001\`, \`imagen-4.0-fast-generate-001\`, \`imagen-4.0-ultra-generate-001\`
   - Gemini (native): \`gemini-3.1-flash-image-preview\`, \`gemini-3-pro-image-preview\`
@@ -289,5 +393,26 @@ export function imagegenActionsPromptSection() {
   - Gemini (Imagen): aspect ratios — \`1:1\` (default), \`3:4\`, \`4:3\`, \`9:16\`, \`16:9\`
   - Gemini (native): size/aspect-ratio params do not apply — omit \`size\` for these models
 - \`quality\` (optional): \`standard\` (default) or \`hd\` — applies to OpenAI dall-e-3 only.
-- \`caption\` (optional): Text message to accompany the image in the channel.`;
+- \`caption\` (optional): Text message to accompany the image in the channel.
+- \`sourceImage\` (optional): Provide a source image for image-to-image editing. **Only supported with native Gemini models** (\`gemini-*\`). Two forms:
+  - **Attachment form** — reference a Discord message attachment:
+    - \`type\` (required): \`"attachment"\`
+    - \`channelId\` (optional): Channel ID of the message containing the image. Defaults to the current channel.
+    - \`messageId\` (optional): Message ID containing the image attachment. Defaults to the current message.
+    - \`attachmentIndex\` (optional): Zero-based index of the attachment to use. Defaults to \`0\` (first attachment).
+    - Example — edit the image from the current message:
+      \`\`\`
+      <discord-action>{"type":"generateImage","prompt":"Make this image look like a watercolor painting","model":"gemini-3.1-flash-image-preview","sourceImage":{"type":"attachment"}}</discord-action>
+      \`\`\`
+    - Example — edit an image from a specific message:
+      \`\`\`
+      <discord-action>{"type":"generateImage","prompt":"Add a sunset sky","model":"gemini-3.1-flash-image-preview","sourceImage":{"type":"attachment","channelId":"123","messageId":"456","attachmentIndex":1}}</discord-action>
+      \`\`\`
+  - **URL form** — provide a public http(s) image URL directly:
+    - \`type\` (required): \`"url"\`
+    - \`url\` (required): A public \`http(s)\` image URL (PNG, JPEG, GIF, or WebP).
+    - Example:
+      \`\`\`
+      <discord-action>{"type":"generateImage","prompt":"Make this photo a pencil sketch","model":"gemini-3.1-flash-image-preview","sourceImage":{"type":"url","url":"https://example.com/photo.jpg"}}</discord-action>
+      \`\`\``;
 }