@roj-ai/sdk 0.1.16 → 0.1.18

package/dist/plugins/uploads/preprocessors/pdf-preprocessor.js

@@ -0,0 +1,274 @@
+/**
+ * PDF Preprocessor
+ *
+ * Dedicated PDF pipeline:
+ *
+ * 1. Text extraction via `pdftotext` (poppler-utils, C++) — ~1 s for a
+ *    3 MB PDF. Replaces markitdown/pdfminer.six (~22 s for the same file)
+ *    because PDFs in practice don't carry the rich markdown structure
+ *    that justifies the slower backend.
+ *
+ * 2. Image extraction via `pdfimages -all` — keeps the original embedded
+ *    format (JPEG stays JPEG) instead of re-encoding everything to PNG
+ *    (~10× faster, much smaller files).
+ *
+ * 3. Text and image extraction run in parallel.
+ *
+ * 4. Images stream into the classifier as soon as `pdfimages` writes them
+ *    to disk — the classifier doesn't wait for the whole extraction to
+ *    finish. A density filter (bytes/pixel) drops alpha masks and overlay
+ *    layers before the vision call.
+ */
+import { dirname } from 'node:path';
+import { Err, Ok } from '../../../lib/utils/result.js';
+import { getImageDimensions, guessImageMime, IMAGE_EXT_RE, MIN_IMAGE_DENSITY_BYTES_PER_PX, MIN_IMAGE_PIXELS, shouldClassifyImage, } from './markitdown-preprocessor.js';
+const PDFTOTEXT_TIMEOUT_MS = 60_000;
+const PDFIMAGES_TIMEOUT_MS = 5 * 60_000;
+const MAX_IMAGES = 20;
+const CLASSIFY_CONCURRENCY = 10;
+const STREAM_POLL_INTERVAL_MS = 250;
+const SUPPORTED_MIME_TYPES = ['application/pdf'];
+export class PdfPreprocessor {
+    name = 'pdf';
+    supportedMimeTypes = SUPPORTED_MIME_TYPES;
+    registry;
+    logger;
+    fs;
+    processRunner;
+    constructor(config) {
+        this.registry = config.registry;
+        this.logger = config.logger;
+        this.fs = config.fs;
+        this.processRunner = config.process;
+    }
+    async process(filePath, mimeType, ctx) {
+        const totalStart = Date.now();
+        this.logger.info('PDF processing started', { filePath });
+        const contentPathResult = ctx.files.realPath('content.md');
+        if (!contentPathResult.ok)
+            return Err(new Error('Failed to resolve content output path'));
+        const imagesDirResult = ctx.files.scoped('images').realPath('');
+        if (!imagesDirResult.ok)
+            return Err(new Error('Failed to resolve images output path'));
+        await this.fs.mkdir(dirname(contentPathResult.value), { recursive: true });
+        await this.fs.mkdir(imagesDirResult.value, { recursive: true });
+        // Run text extraction and image extraction (with streaming classification)
+        // in parallel. They share no state and don't block each other.
+        const [textResult, images] = await Promise.all([
+            this.extractText(filePath, contentPathResult.value),
+            this.extractAndClassifyImages(filePath, imagesDirResult.value, ctx),
+        ]);
+        const markdown = textResult.ok ? textResult.value : '';
+        const derivedPaths = ['content.md'];
+        const imageEntries = [];
+        for (const img of images) {
+            derivedPaths.push(img.relativePath);
+            imageEntries.push(`- ${img.relativePath} — ${img.description}`);
+        }
+        const manifestLines = ['Extracted files:'];
+        manifestLines.push(`- content.md (text, ${markdown.length} chars)`);
+        manifestLines.push(...imageEntries);
+        this.logger.info('PDF processing complete', {
+            filePath,
+            contentLength: markdown.length,
+            imagesClassified: imageEntries.length,
+            totalDurationMs: Date.now() - totalStart,
+        });
+        return Ok({
+            extractedContent: manifestLines.join('\n'),
+            derivedPaths,
+        });
+    }
+    /**
+     * Extract plain text via pdftotext. Writes to content.md verbatim — no
+     * markdown structure to preserve, but the file extension stays .md for
+     * consistency with the markitdown pipeline (downstream consumers expect
+     * "content.md" in the upload directory).
+     *
+     * `-layout` preserves the original visual layout (columns, tables),
+     * which is what users typically expect when looking at PDFs.
+     */
+    async extractText(filePath, outputPath) {
+        const start = Date.now();
+        try {
+            await this.processRunner.execFile('pdftotext', ['-layout', filePath, outputPath], { timeout: PDFTOTEXT_TIMEOUT_MS, maxBuffer: 50 * 1024 * 1024 });
+        }
+        catch (error) {
+            const message = error instanceof Error ? error.message : String(error);
+            this.logger.warn('pdftotext failed', { filePath, durationMs: Date.now() - start, error: message });
+            return Err(new Error(`pdftotext failed: ${message}`));
+        }
+        let text = '';
+        try {
+            text = await this.fs.readFile(outputPath, 'utf-8');
+        }
+        catch {
+            // File missing — pdftotext succeeded but produced no output.
+        }
+        this.logger.info('pdftotext complete', {
+            filePath,
+            durationMs: Date.now() - start,
+            contentLength: text.length,
+        });
+        return Ok(text);
+    }
+    /**
+     * Extract images via pdfimages and classify them as they appear on disk.
+     *
+     * pdfimages writes files atomically per image (open temp, write, rename
+     * to final name), so polling `readdir` is safe — we either see a name or
+     * we don't, never a half-written file.
+     *
+     * Streaming overlaps the extraction tail with the first classification
+     * batches. Hard cap of MAX_IMAGES applies across the *filtered* set: as
+     * soon as MAX_IMAGES images have passed the density filter, further
+     * candidates are stat-checked but not classified.
+     *
+     * `-all` keeps the embedded format (JPEG, JBIG2, JP2). We only classify
+     * those Anthropic vision accepts (PNG/JPEG/GIF/WebP); other formats are
+     * extracted to disk for reference but skipped at the classification step.
+     */
+    async extractAndClassifyImages(filePath, imagesDir, ctx) {
+        const extractStart = Date.now();
+        const seen = new Set();
+        const acceptedQueue = [];
+        const classifyPromises = [];
+        let stopAccepting = false;
+        let droppedByDensity = 0;
+        let skippedUnsupportedExt = 0;
+        // Active classification gate — caps in-flight vision calls.
+        let active = 0;
+        const waiters = [];
+        const acquire = () => new Promise(resolve => {
+            if (active < CLASSIFY_CONCURRENCY) {
+                active++;
+                resolve();
+            }
+            else
+                waiters.push(() => { active++; resolve(); });
+        });
+        const release = () => {
+            active--;
+            const next = waiters.shift();
+            if (next)
+                next();
+        };
+        const classifyOne = async (name) => {
+            await acquire();
+            try {
+                const mime = guessImageMime(name);
+                const fullPath = `${imagesDir}/${name}`;
+                const imageStore = ctx.files.scoped('images');
+                let description = mime;
+                const classifier = this.registry.getForMimeType(mime);
+                if (classifier) {
+                    const result = await classifier.process(fullPath, mime, {
+                        files: ctx.files.scoped(`images/${name}-meta`),
+                    });
+                    if (result.ok && result.value.extractedContent) {
+                        description = result.value.extractedContent;
+                    }
+                }
+                return { relativePath: `images/${name}`, description };
+            }
+            finally {
+                release();
+            }
+        };
+        const inspectAndMaybeClassify = async (name) => {
+            if (seen.has(name) || stopAccepting)
+                return;
+            seen.add(name);
+            if (!IMAGE_EXT_RE.test(name)) {
+                skippedUnsupportedExt++;
+                return;
+            }
+            const fullPath = `${imagesDir}/${name}`;
+            let sizeBytes = 0;
+            try {
+                sizeBytes = (await this.fs.stat(fullPath)).size;
+            }
+            catch {
+                return;
+            }
+            const dims = await getImageDimensions(fullPath, this.processRunner);
+            const hasDims = dims !== null;
+            const passesFilter = hasDims
+                ? shouldClassifyImage({ width: dims.width, height: dims.height, sizeBytes })
+                : sizeBytes >= MIN_IMAGE_PIXELS * MIN_IMAGE_DENSITY_BYTES_PER_PX; // fall back to absolute byte floor
+            if (!passesFilter) {
+                droppedByDensity++;
+                return;
+            }
+            acceptedQueue.push({
+                name,
+                sizeBytes,
+                width: dims?.width ?? 0,
+                height: dims?.height ?? 0,
+            });
+            if (acceptedQueue.length >= MAX_IMAGES) {
+                stopAccepting = true;
+            }
+            classifyPromises.push(classifyOne(name));
+        };
+        // Run pdfimages and a parallel poll loop. The poll calls readdir
+        // periodically and dispatches `inspectAndMaybeClassify` for newly
+        // appeared files; doing it this way avoids fs.watch quirks (some
+        // container filesystems don't deliver events).
+        const pdfimagesPromise = this.processRunner.execFile('pdfimages', ['-all', filePath, `${imagesDir}/img`], { timeout: PDFIMAGES_TIMEOUT_MS, maxBuffer: 1024 * 1024 }).then(() => true).catch((error) => {
+            this.logger.warn('pdfimages failed (will classify any partial output)', {
+                filePath,
+                durationMs: Date.now() - extractStart,
+                error: error instanceof Error ? error.message : String(error),
+            });
+            return false;
+        });
+        let extractionDone = false;
+        const poll = async () => {
+            while (!extractionDone) {
+                await this.scanAndDispatch(imagesDir, inspectAndMaybeClassify);
+                await sleep(STREAM_POLL_INTERVAL_MS);
+            }
+            // Final sweep — pick up anything that landed between the last poll
+            // and pdfimages exiting.
+            await this.scanAndDispatch(imagesDir, inspectAndMaybeClassify);
+        };
+        const pollPromise = poll();
+        const extractSucceeded = await pdfimagesPromise;
+        extractionDone = true;
+        await pollPromise;
+        this.logger.info(extractSucceeded ? 'pdfimages complete' : 'pdfimages failed (partial)', {
+            filePath,
+            durationMs: Date.now() - extractStart,
+            filesEmitted: seen.size,
+            passedFilter: acceptedQueue.length,
+            droppedByDensity,
+            skippedUnsupportedExt,
+        });
+        const classifyStart = Date.now();
+        const settled = await Promise.all(classifyPromises);
+        const images = settled.filter((r) => r !== null);
+        this.logger.info('PDF image classification complete', {
+            filePath,
+            count: images.length,
+            durationMs: Date.now() - classifyStart,
+        });
+        return images;
+    }
+    async scanAndDispatch(dir, handle) {
+        let entries;
+        try {
+            entries = await this.fs.readdir(dir);
+        }
+        catch {
+            return;
+        }
+        // Fire dispatches in parallel — `inspectAndMaybeClassify` is internally
+        // idempotent for already-seen names.
+        await Promise.all(entries.map(handle));
+    }
+}
+function sleep(ms) {
+    return new Promise(resolve => setTimeout(resolve, ms));
+}
+//# sourceMappingURL=pdf-preprocessor.js.map

package/dist/plugins/uploads/preprocessors/pdf-preprocessor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-preprocessor.js","sourceRoot":"","sources":["../../../../src/plugins/uploads/preprocessors/pdf-preprocessor.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;GAoBG;AAEH,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAA;AAEnC,OAAO,EAAE,GAAG,EAAE,EAAE,EAAE,MAAM,uBAAuB,CAAA;AAK/C,OAAO,EACN,kBAAkB,EAClB,cAAc,EACd,YAAY,EACZ,8BAA8B,EAC9B,gBAAgB,EAChB,mBAAmB,GACnB,MAAM,8BAA8B,CAAA;AAErC,MAAM,oBAAoB,GAAG,MAAM,CAAA;AACnC,MAAM,oBAAoB,GAAG,CAAC,GAAG,MAAM,CAAA;AAEvC,MAAM,UAAU,GAAG,EAAE,CAAA;AACrB,MAAM,oBAAoB,GAAG,EAAE,CAAA;AAC/B,MAAM,uBAAuB,GAAG,GAAG,CAAA;AAEnC,MAAM,oBAAoB,GAAG,CAAC,iBAAiB,CAAC,CAAA;AAShD,MAAM,OAAO,eAAe;IAClB,IAAI,GAAG,KAAK,CAAA;IACZ,kBAAkB,GAAG,oBAAoB,CAAA;IAEjC,QAAQ,CAAsB;IAC9B,MAAM,CAAQ;IACd,EAAE,CAAY;IACd,aAAa,CAAe;IAE7C,YAAY,MAA6B;QACxC,IAAI,CAAC,QAAQ,GAAG,MAAM,CAAC,QAAQ,CAAA;QAC/B,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,CAAA;QAC3B,IAAI,CAAC,EAAE,GAAG,MAAM,CAAC,EAAE,CAAA;QACnB,IAAI,CAAC,aAAa,GAAG,MAAM,CAAC,OAAO,CAAA;IACpC,CAAC;IAED,KAAK,CAAC,OAAO,CACZ,QAAgB,EAChB,QAAgB,EAChB,GAAwB;QAExB,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QAC7B,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,wBAAwB,EAAE,EAAE,QAAQ,EAAE,CAAC,CAAA;QAExD,MAAM,iBAAiB,GAAG,GAAG,CAAC,KAAK,CAAC,QAAQ,CAAC,YAAY,CAAC,CAAA;QAC1D,IAAI,CAAC,iBAAiB,CAAC,EAAE;YAAE,OAAO,GAAG,CAAC,IAAI,KAAK,CAAC,uCAAuC,CAAC,CAAC,CAAA;QAEzF,MAAM,eAAe,GAAG,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAA;QAC/D,IAAI,CAAC,eAAe,CAAC,EAAE;YAAE,OAAO,GAAG,CAAC,IAAI,KAAK,CAAC,sCAAsC,CAAC,CAAC,CAAA;QAEtF,MAAM,IAAI,CAAC,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,iBAAiB,CAAC,KAAK,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;QAC1E,MAAM,IAAI,CAAC,EAAE,CAAC,KAAK,CAAC,eAAe,CAAC,KAAK,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAA;QAE/D,2EAA2E;QAC3E,+DAA+D;QAC/D,MAAM,CAAC,UAAU,EAAE,MAAM,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;YAC9C,IAAI,CAAC,WAAW,CAAC,QAAQ,EAAE,iBAAiB,CAAC,KAAK,CAAC;YACnD,IAAI,CAAC,wBAAwB,CAAC,QAAQ,EAAE,eAAe,CAAC,KAAK,EAAE,GAAG,CAAC;SACnE,CAAC,CAAA;QAEF,MAAM,QAAQ,GAAG,UAAU,CAAC,EAAE,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,CAAA;QAEtD,MAAM,YAAY,GAAa,CAAC,YAAY,CAAC,CAAA;QAC7C,MAAM,YAAY,GAAa,EAAE,CAAA;QACjC,KAAK,MAAM,GAAG,IAAI,MAAM,EAAE,CAAC;YAC1B,YAAY,CAAC,IAAI,CAAC,GAAG,CAAC,YAAY,CAAC,CAAA;YACnC,YAAY,CAAC,IAAI,CAAC,KAAK,GAAG,CAAC,YAAY,MAAM,GAAG,CAAC,WAAW,EAAE,CAAC,CAAA;QAChE,CAAC;QAED,MAAM,aAAa,GAAa,CAAC,kBAAkB,CAAC,CAAA;QACpD,aAAa,CAAC,IAAI,CAAC,uBAAuB,QAAQ,CAAC,MAAM,SAAS,CAAC,CAAA;QACnE,aAAa,CAAC,IAAI,CAAC,GAAG,YAAY,CAAC,CAAA;QAEnC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,yBAAyB,EAAE;YAC3C,QAAQ;YACR,aAAa,EAAE,QAAQ,CAAC,MAAM;YAC9B,gBAAgB,EAAE,YAAY,CAAC,MAAM;YACrC,eAAe,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,UAAU;SACxC,CAAC,CAAA;QAEF,OAAO,EAAE,CAAC;YACT,gBAAgB,EAAE,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC;YAC1C,YAAY;SACZ,CAAC,CAAA;IACH,CAAC;IAED;;;;;;;;OAQG;IACK,KAAK,CAAC,WAAW,CAAC,QAAgB,EAAE,UAAkB;QAC7D,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QACxB,IAAI,CAAC;YACJ,MAAM,IAAI,CAAC,aAAa,CAAC,QAAQ,CAChC,WAAW,EACX,CAAC,SAAS,EAAE,QAAQ,EAAE,UAAU,CAAC,EACjC,EAAE,OAAO,EAAE,oBAAoB,EAAE,SAAS,EAAE,EAAE,GAAG,IAAI,GAAG,IAAI,EAAE,CAC9D,CAAA;QACF,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YAChB,MAAM,OAAO,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAA;YACtE,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,kBAAkB,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,KAAK,EAAE,KAAK,EAAE,OAAO,EAAE,CAAC,CAAA;YAClG,OAAO,GAAG,CAAC,IAAI,KAAK,CAAC,qBAAqB,OAAO,EAAE,CAAC,CAAC,CAAA;QACtD,CAAC;QAED,IAAI,IAAI,GAAG,EAAE,CAAA;QACb,IAAI,CAAC;YACJ,IAAI,GAAG,MAAM,IAAI,CAAC,EAAE,CAAC,QAAQ,CAAC,UAAU,EAAE,OAAO,CAAC,CAAA;QACnD,CAAC;QAAC,MAAM,CAAC;YACR,6DAA6D;QAC9D,CAAC;QAED,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,oBAAoB,EAAE;YACtC,QAAQ;YACR,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,KAAK;YAC9B,aAAa,EAAE,IAAI,CAAC,MAAM;SAC1B,CAAC,CAAA;QAEF,OAAO,EAAE,CAAC,IAAI,CAAC,CAAA;IAChB,CAAC;IAED;;;;;;;;;;;;;;;OAeG;IACK,KAAK,CAAC,wBAAwB,CACrC,QAAgB,EAChB,SAAiB,EACjB,GAAwB;QAExB,MAAM,YAAY,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QAC/B,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;QAC9B,MAAM,aAAa,GAA8E,EAAE,CAAA;QACnG,MAAM,gBAAgB,GAAyE,EAAE,CAAA;QACjG,IAAI,aAAa,GAAG,KAAK,CAAA;QACzB,IAAI,gBAAgB,GAAG,CAAC,CAAA;QACxB,IAAI,qBAAqB,GAAG,CAAC,CAAA;QAE7B,4DAA4D;QAC5D,IAAI,MAAM,GAAG,CAAC,CAAA;QACd,MAAM,OAAO,GAAsB,EAAE,CAAA;QACrC,MAAM,OAAO,GAAG,GAAG,EAAE,CAAC,IAAI,OAAO,CAAO,OAAO,CAAC,EAAE;YACjD,IAAI,MAAM,GAAG,oBAAoB,EAAE,CAAC;gBAAC,MAAM,EAAE,CAAC;gBAAC,OAAO,EAAE,CAAA;YAAC,CAAC;;gBACrD,OAAO,CAAC,IAAI,CAAC,GAAG,EAAE,GAAG,MAAM,EAAE,CAAC,CAAC,OAAO,EAAE,CAAA,CAAC,CAAC,CAAC,CAAA;QACjD,CAAC,CAAC,CAAA;QACF,MAAM,OAAO,GAAG,GAAG,EAAE;YACpB,MAAM,EAAE,CAAA;YACR,MAAM,IAAI,GAAG,OAAO,CAAC,KAAK,EAAE,CAAA;YAC5B,IAAI,IAAI;gBAAE,IAAI,EAAE,CAAA;QACjB,CAAC,CAAA;QAED,MAAM,WAAW,GAAG,KAAK,EAAE,IAAY,EAAiE,EAAE;YACzG,MAAM,OAAO,EAAE,CAAA;YACf,IAAI,CAAC;gBACJ,MAAM,IAAI,GAAG,cAAc,CAAC,IAAI,CAAC,CAAA;gBACjC,MAAM,QAAQ,GAAG,GAAG,SAAS,IAAI,IAAI,EAAE,CAAA;gBACvC,MAAM,UAAU,GAAG,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAA;gBAC7C,IAAI,WAAW,GAAG,IAAI,CAAA;gBAEtB,MAAM,UAAU,GAAG,IAAI,CAAC,QAAQ,CAAC,cAAc,CAAC,IAAI,CAAC,CAAA;gBACrD,IAAI,UAAU,EAAE,CAAC;oBAChB,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,EAAE;wBACvD,KAAK,EAAE,GAAG,CAAC,KAAK,CAAC,MAAM,CAAC,UAAU,IAAI,OAAO,CAAC;qBAC9C,CAAC,CAAA;oBACF,IAAI,MAAM,CAAC,EAAE,IAAI,MAAM,CAAC,KAAK,CAAC,gBAAgB,EAAE,CAAC;wBAChD,WAAW,GAAG,MAAM,CAAC,KAAK,CAAC,gBAAgB,CAAA;oBAC5C,CAAC;gBACF,CAAC;gBAED,OAAO,EAAE,YAAY,EAAE,UAAU,IAAI,EAAE,EAAE,WAAW,EAAE,CAAA;YACvD,CAAC;oBAAS,CAAC;gBACV,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC,CAAA;QAED,MAAM,uBAAuB,GAAG,KAAK,EAAE,IAAY,EAAE,EAAE;YACtD,IAAI,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,aAAa;gBAAE,OAAM;YAC3C,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;YAEd,IAAI,CAAC,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;gBAC9B,qBAAqB,EAAE,CAAA;gBACvB,OAAM;YACP,CAAC;YAED,MAAM,QAAQ,GAAG,GAAG,SAAS,IAAI,IAAI,EAAE,CAAA;YACvC,IAAI,SAAS,GAAG,CAAC,CAAA;YACjB,IAAI,CAAC;gBACJ,SAAS,GAAG,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAA;YAChD,CAAC;YAAC,MAAM,CAAC;gBACR,OAAM;YACP,CAAC;YAED,MAAM,IAAI,GAAG,MAAM,kBAAkB,CAAC,QAAQ,EAAE,IAAI,CAAC,aAAa,CAAC,CAAA;YACnE,MAAM,OAAO,GAAG,IAAI,KAAK,IAAI,CAAA;YAC7B,MAAM,YAAY,GAAG,OAAO;gBAC3B,CAAC,CAAC,mBAAmB,CAAC,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,SAAS,EAAE,CAAC;gBAC5E,CAAC,CAAC,SAAS,IAAI,gBAAgB,GAAG,8BAA8B,CAAA,CAAC,mCAAmC;YAErG,IAAI,CAAC,YAAY,EAAE,CAAC;gBACnB,gBAAgB,EAAE,CAAA;gBAClB,OAAM;YACP,CAAC;YAED,aAAa,CAAC,IAAI,CAAC;gBAClB,IAAI;gBACJ,SAAS;gBACT,KAAK,EAAE,IAAI,EAAE,KAAK,IAAI,CAAC;gBACvB,MAAM,EAAE,IAAI,EAAE,MAAM,IAAI,CAAC;aACzB,CAAC,CAAA;YAEF,IAAI,aAAa,CAAC,MAAM,IAAI,UAAU,EAAE,CAAC;gBACxC,aAAa,GAAG,IAAI,CAAA;YACrB,CAAC;YAED,gBAAgB,CAAC,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAA;QACzC,CAAC,CAAA;QAED,iEAAiE;QACjE,kEAAkE;QAClE,iEAAiE;QACjE,+CAA+C;QAC/C,MAAM,gBAAgB,GAAG,IAAI,CAAC,aAAa,CAAC,QAAQ,CACnD,WAAW,EACX,CAAC,MAAM,EAAE,QAAQ,EAAE,GAAG,SAAS,MAAM,CAAC,EACtC,EAAE,OAAO,EAAE,oBAAoB,EAAE,SAAS,EAAE,IAAI,GAAG,IAAI,EAAE,CACzD,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE;YAClC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,qDAAqD,EAAE;gBACvE,QAAQ;gBACR,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,YAAY;gBACrC,KAAK,EAAE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC;aAC7D,CAAC,CAAA;YACF,OAAO,KAAK,CAAA;QACb,CAAC,CAAC,CAAA;QAEF,IAAI,cAAc,GAAG,KAAK,CAAA;QAC1B,MAAM,IAAI,GAAG,KAAK,IAAI,EAAE;YACvB,OAAO,CAAC,cAAc,EAAE,CAAC;gBACxB,MAAM,IAAI,CAAC,eAAe,CAAC,SAAS,EAAE,uBAAuB,CAAC,CAAA;gBAC9D,MAAM,KAAK,CAAC,uBAAuB,CAAC,CAAA;YACrC,CAAC;YACD,mEAAmE;YACnE,yBAAyB;YACzB,MAAM,IAAI,CAAC,eAAe,CAAC,SAAS,EAAE,uBAAuB,CAAC,CAAA;QAC/D,CAAC,CAAA;QAED,MAAM,WAAW,GAAG,IAAI,EAAE,CAAA;QAE1B,MAAM,gBAAgB,GAAG,MAAM,gBAAgB,CAAA;QAC/C,cAAc,GAAG,IAAI,CAAA;QACrB,MAAM,WAAW,CAAA;QAEjB,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC,CAAC,oBAAoB,CAAC,CAAC,CAAC,4BAA4B,EAAE;YACxF,QAAQ;YACR,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,YAAY;YACrC,YAAY,EAAE,IAAI,CAAC,IAAI;YACvB,YAAY,EAAE,aAAa,CAAC,MAAM;YAClC,gBAAgB;YAChB,qBAAqB;SACrB,CAAC,CAAA;QAEF,MAAM,aAAa,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;QAChC,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,gBAAgB,CAAC,CAAA;QACnD,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAsD,EAAE,CAAC,CAAC,KAAK,IAAI,CAAC,CAAA;QAEpG,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,mCAAmC,EAAE;YACrD,QAAQ;YACR,KAAK,EAAE,MAAM,CAAC,MAAM;YACpB,UAAU,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,aAAa;SACtC,CAAC,CAAA;QAEF,OAAO,MAAM,CAAA;IACd,CAAC;IAEO,KAAK,CAAC,eAAe,CAC5B,GAAW,EACX,MAAuC;QAEvC,IAAI,OAAiB,CAAA;QACrB,IAAI,CAAC;YACJ,OAAO,GAAG,MAAM,IAAI,CAAC,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;QACrC,CAAC;QAAC,MAAM,CAAC;YACR,OAAM;QACP,CAAC;QACD,wEAAwE;QACxE,qCAAqC;QACrC,MAAM,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAA;IACvC,CAAC;CACD;AAED,SAAS,KAAK,CAAC,EAAU;IACxB,OAAO,IAAI,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC,UAAU,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC,CAAA;AACvD,CAAC"}

package/dist/transport/http/app.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"app.d.ts","sourceRoot":"","sources":["../../../src/transport/http/app.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,MAAM,CAAA;AAE3B,OAAO,KAAK,EAAE,oBAAoB,EAAE,MAAM,mCAAmC,CAAA;AAC7E,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAA;AAClD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,wCAAwC,CAAA;AAQ5E;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,QAAQ,GAAG;IACpC,cAAc,EAAE,cAAc,CAAA;IAC9B,0FAA0F;IAC1F,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,iHAAiH;IACjH,oBAAoB,CAAC,EAAE,oBAAoB,CAAA;CAC3C,CAAA;AAED;;GAEG;AACH,MAAM,MAAM,MAAM,GAAG;IACpB,SAAS,EAAE;QACV,QAAQ,EAAE,WAAW,CAAA;KACrB,CAAA;CACD,CAAA;AAED;;GAEG;AACH,MAAM,MAAM,UAAU,GAAG,OAAO,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAA;AAEvD;;;GAGG;AACH,wBAAgB,WAAW,CAAC,CAAC,EAAE,UAAU,GAAG,WAAW,CAEtD;AAED;;GAEG;AACH,wBAAgB,SAAS,CAAC,QAAQ,EAAE,WAAW,GAAG,IAAI,CAAC,MAAM,CAAC,CA0E7D"}
+{"version":3,"file":"app.d.ts","sourceRoot":"","sources":["../../../src/transport/http/app.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,MAAM,CAAA;AAG3B,OAAO,KAAK,EAAE,oBAAoB,EAAE,MAAM,mCAAmC,CAAA;AAC7E,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAA;AAClD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,wCAAwC,CAAA;AAQ5E;;GAEG;AACH,MAAM,MAAM,WAAW,GAAG,QAAQ,GAAG;IACpC,cAAc,EAAE,cAAc,CAAA;IAC9B,0FAA0F;IAC1F,UAAU,CAAC,EAAE,MAAM,CAAA;IACnB,iHAAiH;IACjH,oBAAoB,CAAC,EAAE,oBAAoB,CAAA;CAC3C,CAAA;AAED;;GAEG;AACH,MAAM,MAAM,MAAM,GAAG;IACpB,SAAS,EAAE;QACV,QAAQ,EAAE,WAAW,CAAA;KACrB,CAAA;CACD,CAAA;AAED;;GAEG;AACH,MAAM,MAAM,UAAU,GAAG,OAAO,MAAM,EAAE,OAAO,CAAC,MAAM,CAAC,CAAA;AAEvD;;;GAGG;AACH,wBAAgB,WAAW,CAAC,CAAC,EAAE,UAAU,GAAG,WAAW,CAEtD;AAED;;GAEG;AACH,wBAAgB,SAAS,CAAC,QAAQ,EAAE,WAAW,GAAG,IAAI,CAAC,MAAM,CAAC,CA8E7D"}

package/dist/transport/http/app.js

@@ -5,6 +5,7 @@
  */
 import { Hono } from 'hono';
 import { cors } from 'hono/cors';
+import { SDK_VERSION } from '../../info.js';
 import { createBearerAuth } from './middleware/bearer-auth.js';
 import { errorHandler } from './middleware/error-handler.js';
 import { createFileRoutes } from './routes/files.js';
@@ -43,10 +44,14 @@ export function createApp(services) {
     // Activity status for DO polling (protected)
     // Returns lastActivityAt timestamp for the caller to determine if agent is active
     app.get('/status', bearerAuth, async (c) => {
-        const { sessionRuntime } = getServices(c);
+        const { sessionRuntime, config } = getServices(c);
         const stats = await sessionRuntime.getStats();
         return c.json({
             lastActivityAt: stats.lastActivityAt,
+            versions: {
+                sdk: SDK_VERSION,
+                runtime: config.agentRuntime ?? null,
+            },
             stats: {
                 sessionCount: stats.sessionCount,
                 pendingAgents: stats.pendingAgents,

package/dist/transport/http/app.js.map

@@ -1 +1 @@
-{"version":3,"file":"app.js","sourceRoot":"","sources":["../../../src/transport/http/app.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,MAAM,CAAA;AAC3B,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAIhC,OAAO,EAAE,gBAAgB,EAAE,MAAM,6BAA6B,CAAA;AAC9D,OAAO,EAAE,YAAY,EAAE,MAAM,+BAA+B,CAAA;AAC5D,OAAO,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAA;AACpD,OAAO,EAAE,oBAAoB,EAAE,MAAM,uBAAuB,CAAA;AAC5D,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAA;AACjD,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAA;AA2BvD;;;GAGG;AACH,MAAM,UAAU,WAAW,CAAC,CAAa;IACxC,OAAO,CAAC,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;AACzB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,SAAS,CAAC,QAAqB;IAC9C,MAAM,GAAG,GAAG,IAAI,IAAI,EAAU,CAAA;IAE9B,aAAa;IACb,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;IACpB,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,EAAE;QAC9B,CAAC,CAAC,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAA;QAC3B,MAAM,IAAI,EAAE,CAAA;IACb,CAAC,CAAC,CAAA;IAEF,gBAAgB;IAChB,GAAG,CAAC,OAAO,CAAC,YAAY,CAAC,CAAA;IAEzB,2CAA2C;IAC3C,GAAG,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC,EAAE,EAAE;QACxB,OAAO,CAAC,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,IAAI;YACZ,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE;SACrB,CAAC,CAAA;IACH,CAAC,CAAC,CAAA;IAEF,mCAAmC;IACnC,MAAM,UAAU,GAAG,gBAAgB,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAA;IAExD,6CAA6C;IAC7C,kFAAkF;IAClF,GAAG,CAAC,GAAG,CAAC,SAAS,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC,EAAE,EAAE;QAC1C,MAAM,EAAE,cAAc,EAAE,GAAG,WAAW,CAAC,CAAC,CAAC,CAAA;QACzC,MAAM,KAAK,GAAG,MAAM,cAAc,CAAC,QAAQ,EAAE,CAAA;QAE7C,OAAO,CAAC,CAAC,IAAI,CAAC;YACb,cAAc,EAAE,KAAK,CAAC,cAAc;YACpC,KAAK,EAAE;gBACN,YAAY,EAAE,KAAK,CAAC,YAAY;gBAChC,aAAa,EAAE,KAAK,CAAC,aAAa;gBAClC,gBAAgB,EAAE,KAAK,CAAC,gBAAgB;aACxC;YACD,QAAQ,EAAE,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;gBAClC,EAAE,EAAE,CAAC,CAAC,EAAE;gBACR,QAAQ,EAAE,CAAC,CAAC,QAAQ;gBACpB,MAAM,EAAE,CAAC,CAAC,MAAM;gBAChB,OAAO,EAAE,CAAC,CAAC,OAAO;aAClB,CAAC,CAAC;YACH,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE;SACrB,CAAC,CAAA;IACH,CAAC,CAAC,CAAA;IAEF,mBAAmB;IACnB,MAAM,SAAS,GAAG,eAAe,EAAE,CAAA;IACnC,MAAM,YAAY,GAAG,kBAAkB,EAAE,CAAA;IACzC,MAAM,cAAc,GAAG,oBAAoB,EAAE,CAAA;IAC7C,MAAM,UAAU,GAAG,gBAAgB,EAAE,CAAA;IAErC,GAAG,CAAC,GAAG,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAA;IAC7B,GAAG,CAAC,GAAG,CAAC,aAAa,EAAE,UAAU,CAAC,CAAA;IAClC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,SAAS,CAAC,CAAA;IAC5B,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,YAAY,CAAC,CAAA;IACpC,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,cAAc,CAAC,CAAA;IACtC,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,UAAU,CAAC,CAAA;IAElC,cAAc;IACd,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE;QAClB,OAAO,CAAC,CAAC,IAAI,CACZ;YACC,KAAK,EAAE;gBACN,IAAI,EAAE,WAAW;gBACjB,OAAO,EAAE,oBAAoB,CAAC,CAAC,GAAG,CAAC,MAAM,IAAI,CAAC,CAAC,GAAG,CAAC,IAAI,EAAE;aACzD;SACD,EACD,GAAG,CACH,CAAA;IACF,CAAC,CAAC,CAAA;IAEF,OAAO,GAAG,CAAA;AACX,CAAC"}
+{"version":3,"file":"app.js","sourceRoot":"","sources":["../../../src/transport/http/app.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,MAAM,MAAM,CAAA;AAC3B,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAA;AAChC,OAAO,EAAE,WAAW,EAAE,MAAM,WAAW,CAAA;AAIvC,OAAO,EAAE,gBAAgB,EAAE,MAAM,6BAA6B,CAAA;AAC9D,OAAO,EAAE,YAAY,EAAE,MAAM,+BAA+B,CAAA;AAC5D,OAAO,EAAE,gBAAgB,EAAE,MAAM,mBAAmB,CAAA;AACpD,OAAO,EAAE,oBAAoB,EAAE,MAAM,uBAAuB,CAAA;AAC5D,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAA;AACjD,OAAO,EAAE,kBAAkB,EAAE,MAAM,oBAAoB,CAAA;AA2BvD;;;GAGG;AACH,MAAM,UAAU,WAAW,CAAC,CAAa;IACxC,OAAO,CAAC,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;AACzB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,SAAS,CAAC,QAAqB;IAC9C,MAAM,GAAG,GAAG,IAAI,IAAI,EAAU,CAAA;IAE9B,aAAa;IACb,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;IACpB,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,EAAE,CAAC,EAAE,IAAI,EAAE,EAAE;QAC9B,CAAC,CAAC,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAA;QAC3B,MAAM,IAAI,EAAE,CAAA;IACb,CAAC,CAAC,CAAA;IAEF,gBAAgB;IAChB,GAAG,CAAC,OAAO,CAAC,YAAY,CAAC,CAAA;IAEzB,2CAA2C;IAC3C,GAAG,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC,EAAE,EAAE;QACxB,OAAO,CAAC,CAAC,IAAI,CAAC;YACb,MAAM,EAAE,IAAI;YACZ,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE;SACrB,CAAC,CAAA;IACH,CAAC,CAAC,CAAA;IAEF,mCAAmC;IACnC,MAAM,UAAU,GAAG,gBAAgB,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAA;IAExD,6CAA6C;IAC7C,kFAAkF;IAClF,GAAG,CAAC,GAAG,CAAC,SAAS,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC,EAAE,EAAE;QAC1C,MAAM,EAAE,cAAc,EAAE,MAAM,EAAE,GAAG,WAAW,CAAC,CAAC,CAAC,CAAA;QACjD,MAAM,KAAK,GAAG,MAAM,cAAc,CAAC,QAAQ,EAAE,CAAA;QAE7C,OAAO,CAAC,CAAC,IAAI,CAAC;YACb,cAAc,EAAE,KAAK,CAAC,cAAc;YACpC,QAAQ,EAAE;gBACT,GAAG,EAAE,WAAW;gBAChB,OAAO,EAAE,MAAM,CAAC,YAAY,IAAI,IAAI;aACpC;YACD,KAAK,EAAE;gBACN,YAAY,EAAE,KAAK,CAAC,YAAY;gBAChC,aAAa,EAAE,KAAK,CAAC,aAAa;gBAClC,gBAAgB,EAAE,KAAK,CAAC,gBAAgB;aACxC;YACD,QAAQ,EAAE,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;gBAClC,EAAE,EAAE,CAAC,CAAC,EAAE;gBACR,QAAQ,EAAE,CAAC,CAAC,QAAQ;gBACpB,MAAM,EAAE,CAAC,CAAC,MAAM;gBAChB,OAAO,EAAE,CAAC,CAAC,OAAO;aAClB,CAAC,CAAC;YACH,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE;SACrB,CAAC,CAAA;IACH,CAAC,CAAC,CAAA;IAEF,mBAAmB;IACnB,MAAM,SAAS,GAAG,eAAe,EAAE,CAAA;IACnC,MAAM,YAAY,GAAG,kBAAkB,EAAE,CAAA;IACzC,MAAM,cAAc,GAAG,oBAAoB,EAAE,CAAA;IAC7C,MAAM,UAAU,GAAG,gBAAgB,EAAE,CAAA;IAErC,GAAG,CAAC,GAAG,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAA;IAC7B,GAAG,CAAC,GAAG,CAAC,aAAa,EAAE,UAAU,CAAC,CAAA;IAClC,GAAG,CAAC,KAAK,CAAC,MAAM,EAAE,SAAS,CAAC,CAAA;IAC5B,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,YAAY,CAAC,CAAA;IACpC,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,cAAc,CAAC,CAAA;IACtC,GAAG,CAAC,KAAK,CAAC,WAAW,EAAE,UAAU,CAAC,CAAA;IAElC,cAAc;IACd,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE;QAClB,OAAO,CAAC,CAAC,IAAI,CACZ;YACC,KAAK,EAAE;gBACN,IAAI,EAAE,WAAW;gBACjB,OAAO,EAAE,oBAAoB,CAAC,CAAC,GAAG,CAAC,MAAM,IAAI,CAAC,CAAC,GAAG,CAAC,IAAI,EAAE;aACzD;SACD,EACD,GAAG,CACH,CAAA;IACF,CAAC,CAAC,CAAA;IAEF,OAAO,GAAG,CAAA;AACX,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@roj-ai/sdk",
-	"version": "0.1.16",
+	"version": "0.1.18",
 	"type": "module",
 	"main": "./dist/index.js",
 	"exports": {
@@ -135,7 +135,7 @@
 		"type-check": "tsc --noEmit"
 	},
 	"dependencies": {
-		"@roj-ai/transport": "^0.1.16",
+		"@roj-ai/transport": "^0.1.18",
 		"@hono/zod-validator": "0.7.6",
 		"hono": "4.12.5",
 		"ignore": "7.0.5",

package/src/bootstrap.ts

@@ -18,7 +18,7 @@ import type { RoutableLLMProvider } from '~/core/llm/routing-provider.js'
 import type { Preset } from '~/core/preset/index.js'
 import type { Platform } from '~/platform/index.js'
 import { PreprocessorRegistry } from '~/plugins/uploads/preprocessor.js'
-import { ImageClassifierPreprocessor, MarkitdownPreprocessor, ZipPreprocessor } from '~/plugins/uploads/preprocessors/index.js'
+import { ImageClassifierPreprocessor, MarkitdownPreprocessor, PdfPreprocessor, ZipPreprocessor } from '~/plugins/uploads/preprocessors/index.js'
 import type { Config } from './config.js'
 import { SessionFileStore } from './core/file-store/file-store.js'
 import type { SessionManager } from './core/sessions/session-manager.js'
@@ -121,7 +121,17 @@ export function bootstrap(config: Config, userConfig: RojConfig, platform: Platf
 
 	const preprocessorRegistry = new PreprocessorRegistry()
 	const imageClassifierGate = new Semaphore(config.imageClassifierConcurrency ?? 10)
-	preprocessorRegistry.register(new ImageClassifierPreprocessor({ llmProvider, logger, fs: platform.fs, gate: imageClassifierGate }))
+	// Dedicated resizer for the classification path: separate from the LLM
+	// provider's general-purpose ImageProcessor (which keeps a higher
+	// maxDimension for agent file-inspection tool calls). The classifier
+	// hands each image to the resizer with its own 1024px override.
+	const classifierImageResizer = new VipsImageResizer({ fs: platform.fs, process: platform.process, tmpDir: platform.tmpDir })
+	preprocessorRegistry.register(new ImageClassifierPreprocessor({ llmProvider, logger, fs: platform.fs, gate: imageClassifierGate, imageResizer: classifierImageResizer }))
+	// PdfPreprocessor must come before MarkitdownPreprocessor — both could
+	// match `application/pdf` in principle, but the registry uses first-hit
+	// and PdfPreprocessor's `pdftotext + pdfimages -all + streaming` pipeline
+	// is dramatically faster than markitdown's pdfminer.six backend.
+	preprocessorRegistry.register(new PdfPreprocessor({ registry: preprocessorRegistry, logger, fs: platform.fs, process: platform.process }))
 	preprocessorRegistry.register(new MarkitdownPreprocessor({ registry: preprocessorRegistry, logger, fs: platform.fs, process: platform.process }))
 	preprocessorRegistry.register(new ZipPreprocessor({ registry: preprocessorRegistry, logger, process: platform.process }))
 

package/src/config.ts

@@ -33,6 +33,19 @@ export interface Config {
 	/** Max concurrent vision LLM calls when classifying uploaded images. Default 10. */
 	imageClassifierConcurrency?: number
 
+	/**
+	 * Identity of the application embedding this SDK. Reported via `/status`
+	 * so platform health-checks can surface "what's actually running" in debug
+	 * tooling alongside the SDK's own version.
+	 *
+	 * sandbox-runtime sets this from its own package.json; other embedders
+	 * (e.g. standalone-server, custom bundles) can supply their own.
+	 */
+	agentRuntime?: {
+		name: string
+		version: string
+	}
+
 	// Logging
 	logLevel: LogLevel
 	logFormat: 'console' | 'json'

package/src/core/image/types.ts

@@ -3,6 +3,8 @@ import type { ToolResultContent } from '~/core/llm/llm-log-types.js'
 
 export interface ImageResizeOptions {
 	maxFileSizeBytes?: number
+	/** Override the resizer's default max dimension (long side, px). */
+	maxDimension?: number
 }
 
 export interface ImageResizer {

package/src/core/image/vips-resizer.ts

@@ -24,9 +24,10 @@ export class VipsImageResizer implements ImageResizer {
 	}
 
 	async resize(filePath: string, mimeType: string, options?: ImageResizeOptions): Promise<ImageResizeResult> {
+		const effectiveMaxDimension = options?.maxDimension ?? this.maxDimension
 		try {
 			// Step 1: Dimension resize if needed
-			const result = await this.dimensionResize(filePath, mimeType)
+			const result = await this.dimensionResize(filePath, mimeType, effectiveMaxDimension)
 
 			// Step 2: If no size constraint, done
 			if (!options?.maxFileSizeBytes) return result
@@ -40,7 +41,7 @@ export class VipsImageResizer implements ImageResizer {
 				await this.fs.unlink(result.tempFile).catch(() => {})
 			}
 
-			return await this.compressToFit(filePath, options.maxFileSizeBytes)
+			return await this.compressToFit(filePath, options.maxFileSizeBytes, effectiveMaxDimension)
 		} catch (e) {
 			console.warn('[image-resize] failed, using original image:', e instanceof Error ? e.message : e)
 			return { path: filePath, mimeType }
@@ -57,9 +58,9 @@ export class VipsImageResizer implements ImageResizer {
 		return { width, height }
 	}
 
-	private async dimensionResize(filePath: string, mimeType: string): Promise<ImageResizeResult> {
+	private async dimensionResize(filePath: string, mimeType: string, maxDimension: number): Promise<ImageResizeResult> {
 		const dims = await this.getImageDimensions(filePath)
-		const needsResize = dims !== null && (dims.width > this.maxDimension || dims.height > this.maxDimension)
+		const needsResize = dims !== null && (dims.width > maxDimension || dims.height > maxDimension)
 
 		// JPEGs within dimension limits pass through unchanged
 		if (mimeType === 'image/jpeg' && !needsResize) {
@@ -73,20 +74,20 @@ export class VipsImageResizer implements ImageResizer {
 		await this.process.execFile('vipsthumbnail', [
 			filePath,
 			'--size',
-			`${this.maxDimension}x${this.maxDimension}`,
+			`${maxDimension}x${maxDimension}`,
 			'-o',
 			outputPath,
 		], { timeout: 30_000 })
 		return { path: outputPath, mimeType: 'image/jpeg', tempFile: outputPath }
 	}
 
-	private async compressToFit(filePath: string, maxFileSizeBytes: number): Promise<ImageResizeResult> {
-		const halfDim = Math.floor(this.maxDimension / 2)
+	private async compressToFit(filePath: string, maxFileSizeBytes: number, maxDimension: number): Promise<ImageResizeResult> {
+		const halfDim = Math.floor(maxDimension / 2)
 		const attempts = [
-			{ dimension: this.maxDimension, quality: 85 },
-			{ dimension: this.maxDimension, quality: 70 },
-			{ dimension: this.maxDimension, quality: 50 },
-			{ dimension: this.maxDimension, quality: 30 },
+			{ dimension: maxDimension, quality: 85 },
+			{ dimension: maxDimension, quality: 70 },
+			{ dimension: maxDimension, quality: 50 },
+			{ dimension: maxDimension, quality: 30 },
 			{ dimension: halfDim, quality: 70 },
 			{ dimension: halfDim, quality: 50 },
 			{ dimension: halfDim, quality: 30 },

package/src/info.ts

@@ -0,0 +1,9 @@
+/**
+ * Build-time SDK metadata. Read from this package's own package.json so the
+ * value tracks releases automatically — no risk of a forgotten string update
+ * after a version bump.
+ */
+
+import pkg from '../package.json' with { type: 'json' }
+
+export const SDK_VERSION = pkg.version as string

package/src/plugins/uploads/preprocessors/image-classifier.ts

@@ -5,6 +5,7 @@
  * Falls back to basic metadata if vision is not available.
  */
 
+import type { ImageResizer } from '~/core/image/types.js'
 import type { LLMProvider } from '~/core/llm/provider.js'
 import { ModelId } from '~/core/llm/schema.js'
 import type { Semaphore } from '~/lib/utils/concurrency.js'
@@ -14,6 +15,15 @@ import type { FileSystem } from '~/platform/fs.js'
 import type { Logger } from '../../../lib/logger/logger.js'
 import type { Preprocessor, PreprocessorContext, PreprocessorResult } from '../preprocessor.js'
 
+/**
+ * Anthropic vision API internally downsamples images to ~1568px long side.
+ * Anything larger just wastes bandwidth and LLM tokens. For 1–2 sentence
+ * descriptions, 1024px is more than enough detail.
+ */
+const CLASSIFY_MAX_DIMENSION = 1024
+/** Hard cap to keep base64 payloads small (LLM still accepts up to 5MB). */
+const CLASSIFY_MAX_FILE_SIZE_BYTES = 1 * 1024 * 1024
+
 // ============================================================================
 // Configuration
 // ============================================================================
@@ -27,6 +37,14 @@ export interface ImageClassifierConfig {
 	logger: Logger
 	/** FileSystem adapter (for checking + reading image files) */
 	fs: FileSystem
+	/**
+	 * Optional resizer. When provided, classifier downscales images to
+	 * ~1024px before sending to the vision LLM. Skipping this is fine for
+	 * tests; in production it dramatically cuts payload size and token cost
+	 * (brand PDFs embed 2000–3000px JPEGs the model would otherwise see at
+	 * full resolution).
+	 */
+	imageResizer?: ImageResizer
 	/** Whether to skip vision and just return metadata */
 	skipVision?: boolean
 	/**
@@ -54,6 +72,7 @@ export class ImageClassifierPreprocessor implements Preprocessor {
 	private readonly visionModel: ModelId
 	private readonly logger: Logger
 	private readonly fs: FileSystem
+	private readonly imageResizer: ImageResizer | undefined
 	private readonly skipVision: boolean
 	private readonly gate: Semaphore | undefined
 
@@ -62,6 +81,7 @@ export class ImageClassifierPreprocessor implements Preprocessor {
 		this.visionModel = config.visionModel ? ModelId(config.visionModel) : ModelId('anthropic/claude-haiku-4.5')
 		this.logger = config.logger
 		this.fs = config.fs
+		this.imageResizer = config.imageResizer
 		this.skipVision = config.skipVision ?? false
 		this.gate = config.gate
 	}
@@ -146,36 +166,41 @@ export class ImageClassifierPreprocessor implements Preprocessor {
 		mimeType: string,
 	): Promise<string | null> {
 		try {
-			// Use file:// URL - resolved to base64 lazily in LLM provider
-			const inferenceCall = () => this.llmProvider.inference({
-				model: this.visionModel,
-				systemPrompt: 'You are an image description assistant. Describe images concisely in 1-2 sentences.',
-				messages: [
-					{
-						role: 'user',
-						content: [
-							{
-								type: 'text',
-								text: 'Please describe this image concisely in 1-2 sentences. Focus on the main subject and any text visible.',
-							},
-							{
-								type: 'image_url',
-								imageUrl: { url: `file://${filePath}` },
-							},
-						],
-					},
-				],
-				maxTokens: 200,
-				temperature: 0.3,
-			})
+			const { url: imageUrl, cleanup } = await this.prepareImageUrl(filePath, mimeType)
 
-			const result = await (this.gate ? this.gate.run(inferenceCall) : inferenceCall())
+			try {
+				const inferenceCall = () => this.llmProvider.inference({
+					model: this.visionModel,
+					systemPrompt: 'You are an image description assistant. Describe images concisely in 1-2 sentences.',
+					messages: [
+						{
+							role: 'user',
+							content: [
+								{
+									type: 'text',
+									text: 'Please describe this image concisely in 1-2 sentences. Focus on the main subject and any text visible.',
+								},
+								{
+									type: 'image_url',
+									imageUrl: { url: imageUrl },
+								},
+							],
+						},
+					],
+					maxTokens: 200,
+					temperature: 0.3,
+				})
 
-			if (result.ok && result.value.content) {
-				return result.value.content.trim()
-			}
+				const result = await (this.gate ? this.gate.run(inferenceCall) : inferenceCall())
 
-			return null
+				if (result.ok && result.value.content) {
+					return result.value.content.trim()
+				}
+
+				return null
+			} finally {
+				await cleanup()
+			}
 		} catch (error) {
 			this.logger.warn('Vision inference failed', {
 				error: error instanceof Error ? error.message : String(error),
@@ -184,6 +209,47 @@ export class ImageClassifierPreprocessor implements Preprocessor {
 		}
 	}
 
+	/**
+	 * Pre-resize the image for vision classification.
+	 *
+	 * Returns either a `data:` URL with the resized JPEG (when a resizer is
+	 * available) or a `file://` URL fallback (LLM provider will resolve it via
+	 * the global ImageProcessor, with its bigger maxDimension default).
+	 *
+	 * Cleanup removes any temp file produced by the resizer.
+	 */
+	private async prepareImageUrl(
+		filePath: string,
+		mimeType: string,
+	): Promise<{ url: string; cleanup: () => Promise<void> }> {
+		if (!this.imageResizer) {
+			return { url: `file://${filePath}`, cleanup: async () => {} }
+		}
+
+		try {
+			const resized = await this.imageResizer.resize(filePath, mimeType, {
+				maxDimension: CLASSIFY_MAX_DIMENSION,
+				maxFileSizeBytes: CLASSIFY_MAX_FILE_SIZE_BYTES,
+			})
+			const buffer = await this.fs.readFile(resized.path)
+			const base64 = buffer.toString('base64')
+			return {
+				url: `data:${resized.mimeType};base64,${base64}`,
+				cleanup: async () => {
+					if (resized.tempFile) {
+						await this.fs.unlink(resized.tempFile).catch(() => {})
+					}
+				},
+			}
+		} catch (error) {
+			this.logger.warn('Pre-resize for classification failed, falling back to file:// URL', {
+				filePath,
+				error: error instanceof Error ? error.message : String(error),
+			})
+			return { url: `file://${filePath}`, cleanup: async () => {} }
+		}
+	}
+
 	private formatSize(bytes: number): string {
 		if (bytes < 1024) return `${bytes}B`
 		if (bytes < 1024 * 1024) return `${(bytes / 1024).toFixed(1)}KB`

package/src/plugins/uploads/preprocessors/index.ts

@@ -4,4 +4,5 @@
 
 export { createImageClassifierPreprocessor, type ImageClassifierConfig, ImageClassifierPreprocessor } from './image-classifier.js'
 export { MarkitdownPreprocessor, type MarkitdownPreprocessorConfig } from './markitdown-preprocessor.js'
+export { PdfPreprocessor, type PdfPreprocessorConfig } from './pdf-preprocessor.js'
 export { ZipPreprocessor, type ZipPreprocessorConfig } from './zip-preprocessor.js'