rhombus-node-mcp 0.1.45 → 0.1.46

package/dist/api/camera-tool-api.js

@@ -1,70 +1,65 @@
 import { getLogger } from "../logger.js";
-import { constructRequestHeaders, postApi } from "../network/network.js";
+import { postApi } from "../network/network.js";
 import { removeNullFields } from "../util.js";
 const logger = getLogger("camera-tool");
 const MILLISECONDS_PER_DAY = 24 * 60 * 60 * 1000;
-export async function getImageForCameraAtTime(cameraUuid, timestampMs, requestModifiers, sessionId) {
+/** Convert a 0-100 percentage to permyriad (0-10000), clamped to the valid range. */
+function pctToPermyriad(pct) {
+    return Math.round(Math.max(0, Math.min(100, pct)) * 100);
+}
+export async function getImageForCameraAtTime(cameraUuid, timestampMs, requestModifiers, sessionId, options) {
+    // biome-ignore lint/suspicious/noExplicitAny: request body is a loose JSON shape
     const body = {
-        cameraUuid: cameraUuid,
-        downscaleFactor: 10,
+        cameraUuid,
+        timestampMs,
+        downscaleFactor: options?.downscaleFactor ?? 10,
         jpgQuality: 70,
-        timestampMs: timestampMs,
     };
-    logger.debug(`Getting frameUri from UUID: ${cameraUuid} at timestampMs: ${timestampMs}`);
-    let frameUri;
-    const base64Image = await postApi({
-        route: "/video/getExactFrameUri",
+    const crop = options?.crop;
+    const hasCrop = crop != null &&
+        (crop.x != null || crop.y != null || crop.width != null || crop.height != null);
+    const resolvedCrop = hasCrop
+        ? {
+            x: crop.x ?? 0,
+            y: crop.y ?? 0,
+            width: crop.width ?? 100,
+            height: crop.height ?? 100,
+        }
+        : null;
+    if (resolvedCrop) {
+        body.permyriadCropX = pctToPermyriad(resolvedCrop.x);
+        body.permyriadCropY = pctToPermyriad(resolvedCrop.y);
+        body.permyriadCropWidth = pctToPermyriad(resolvedCrop.width);
+        body.permyriadCropHeight = pctToPermyriad(resolvedCrop.height);
+    }
+    logger.debug(`Getting exact frame data for UUID: ${cameraUuid} at timestampMs: ${timestampMs}` +
+        (hasCrop ? ` (cropped)` : ``));
+    const res = await postApi({
+        route: "/video/getExactFrameData",
         body,
         modifiers: requestModifiers,
         sessionId,
-    }).then(async (res) => {
-        logger.debug(`Received frameUri ${res.frameUri}`);
-        if (!res.frameUri) {
-            throw new Error("No frameUri given. Maybe camera does not support it.");
-        }
-        // construct request headers
-        const { url: _frameUri, requestHeaders } = constructRequestHeaders(res.frameUri, requestModifiers, sessionId);
-        frameUri = _frameUri;
-        // remove content type
-        delete requestHeaders["Content-Type"];
-        delete requestHeaders["accept"];
-        logger.debug(`Fetching with headers\n${JSON.stringify(requestHeaders, null, 2)}`);
-        return await fetch(frameUri, {
-            method: "GET",
-            headers: requestHeaders,
-        }).then(async (res) => {
-            if (!res.ok) {
-                logger.error(`Failed to fetch image (HTTP ${res.status}): ${await res.text()}`);
-                logger.error(res);
-                return res.status === 404 ? 404 : null;
-            }
-            const arrayBuffer = await res.arrayBuffer();
-            const buffer = Buffer.from(arrayBuffer);
-            const base64 = buffer.toString("base64");
-            logger.debug(`Received image base64:\n ${base64}`);
-            return base64;
-        });
     });
-    if (!base64Image) {
-        return {
-            success: false,
-            status: "failed to fetch image",
-        };
-    }
-    if (base64Image === 404) {
+    if (res.error || !res.frameData) {
+        logger.error(`getExactFrameData failed: ${JSON.stringify({ error: res.error, errorMsg: res.errorMsg, status: res.status, hasFrameData: !!res.frameData })}`);
         return {
             success: false,
             status: "failed to fetch image",
-            message: "Camera snapshot unavailable: this camera has no recorded video (VOD) at the requested time. " +
-                "It may not have any footage stored, or the requested timestamp may be outside its retention window.",
+            // Prefer a structured business error from the endpoint, then any transport-level
+            // status set by postApi (e.g. permission/HTTP errors), and only fall back to the
+            // no-VOD explanation when no error detail is available.
+            message: res.errorMsg ??
+                res.status ??
+                "Camera snapshot unavailable: this camera may have no recorded video (VOD) at the requested time. " +
+                    "It may not have any footage stored, or the requested timestamp may be outside its retention window.",
         };
     }
     return {
         success: true,
         status: "successfully fetched image",
-        frameUri: frameUri ?? "",
         imageType: "base64",
-        imageData: base64Image,
+        imageData: res.frameData,
+        crop: resolvedCrop,
     };
 }
 async function getCameraStorageData(cameraUuid, requestModifiers, sessionId) {

package/dist/tools-console/camera-tool.js

@@ -21,8 +21,8 @@ This tool captures and returns a real-time snapshot from a designated security c
 The image reflects the current scene in the camera's field of view and serves as a contextual
 input source for downstream tasks such as object recognition, anomaly detection, incident investigation,
 or situational assessment. When invoked, the tool provides the following: 
-- Visual Scene Capture: A high-resolution image of what the camera is actively observing, including people, vehicles, license plates, and any detectable objects.  
-- The frameUri that was used to fetch the image. It may be useful to show the user this image as well through the frameUri.
+- Visual Scene Capture: A high-resolution image of what the camera is actively observing, including people, vehicles, license plates, and any detectable objects.
+- Optional zoom: pass cropX, cropY, cropWidth, cropHeight (each a percentage 0-100, origin at the top-left) to return only a sub-region of the frame so you can inspect a detail (e.g. a license plate or a doorway) more closely. Omit them for the full frame. When zooming into a small crop, pass a smaller downscaleFactor (e.g. 1-3) to preserve detail.
 
 What follows is a description of the behavior of this tool given the requestType "get-settings"
 
@@ -56,7 +56,7 @@ Examples that REQUIRE the automatic snapshot flow:
 const logger = getLogger("camera-tool");
 const TOOL_ARGS = BASE_TOOL_ARGS;
 const TOOL_HANDLER = async (args, extra) => {
-    const { cameraUuid, timestampISO, requestType } = args;
+    const { cameraUuid, timestampISO, requestType, cropX, cropY, cropWidth, cropHeight, downscaleFactor } = args;
     if (!cameraUuid) {
         return {
             content: [
@@ -76,13 +76,21 @@ const TOOL_HANDLER = async (args, extra) => {
     const { requestModifiers, sessionId } = extractFromToolExtra(extra);
     switch (requestType) {
         case "image":
-            response = await getImageForCameraAtTime(cameraUuid, timestampMs, requestModifiers, sessionId);
+            response = await getImageForCameraAtTime(cameraUuid, timestampMs, requestModifiers, sessionId, {
+                crop: {
+                    x: cropX ?? null,
+                    y: cropY ?? null,
+                    width: cropWidth ?? null,
+                    height: cropHeight ?? null,
+                },
+                downscaleFactor: downscaleFactor ?? null,
+            });
             if (!response.success || !response.imageData) {
                 return {
                     content: [{ type: "text", text: JSON.stringify(response) }],
                 };
             }
-            logger.debug(`Received image response:\n ${JSON.stringify(response)}`);
+            logger.debug(`Received image response (base64 length ${response.imageData.length})`);
             return {
                 content: [
                     {
@@ -97,7 +105,7 @@ const TOOL_HANDLER = async (args, extra) => {
                             status: "image-attached",
                             cameraUuid,
                             timestampMs,
-                            frameUri: response.frameUri,
+                            cropApplied: response.crop ?? null,
                         }),
                     },
                 ],

package/dist/types/camera-tool-types.js

@@ -221,5 +221,39 @@ export const BASE_TOOL_ARGS = {
       the timestamp for the image. This will default to 5 minutes before the current time. You can also call time-tool to parse the user's time description.
       ` + ISOTimestampFormatDescription),
     requestType: z.enum(["image", "get-settings", "get-media-uris", "get-ai-thresholds"]),
+    cropX: z
+        .number()
+        .min(0)
+        .max(100)
+        .nullable()
+        .optional()
+        .describe("For requestType 'image' only. Left edge of the crop region as a percentage (0-100) of image width, measured from the left. 0 = left edge. Omit for full frame."),
+    cropY: z
+        .number()
+        .min(0)
+        .max(100)
+        .nullable()
+        .optional()
+        .describe("For requestType 'image' only. Top edge of the crop region as a percentage (0-100) of image height, measured from the top. 0 = top edge. Omit for full frame."),
+    cropWidth: z
+        .number()
+        .min(0)
+        .max(100)
+        .nullable()
+        .optional()
+        .describe("For requestType 'image' only. Width of the crop region as a percentage (0-100) of image width. Omit for full frame (100)."),
+    cropHeight: z
+        .number()
+        .min(0)
+        .max(100)
+        .nullable()
+        .optional()
+        .describe("For requestType 'image' only. Height of the crop region as a percentage (0-100) of image height. Omit for full frame (100)."),
+    downscaleFactor: z
+        .number()
+        .min(1)
+        .nullable()
+        .optional()
+        .describe("For requestType 'image' only. Ratio to shrink the image pixels by (default 10). Use a smaller value (e.g. 1-3) when zooming into a crop to keep detail."),
 };
 const BASE_TOOL_ARGS_SCHEMA = z.object(BASE_TOOL_ARGS);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "rhombus-node-mcp",
-  "version": "0.1.45",
+  "version": "0.1.46",
   "description": "MCP server for Rhombus API",
   "keywords": [
     "ai",