sparkecoder 0.1.79 → 0.1.81

package/dist/skills/default/browser.md

@@ -81,17 +81,26 @@ agent-browser snapshot -i                  # Interactive elements only
 ```
 
 ### Screenshots
+
+Screenshots are saved relative to the **working directory**. Always create the directory first and verify `.gitignore` includes the artifacts folder:
+
 ```bash
 mkdir -p ".sparkecode-artifacts/screenshots"
+grep -q "sparkecode-artifacts" .gitignore 2>/dev/null || echo ".sparkecode-artifacts/" >> .gitignore
+```
+
+Take screenshots using relative paths from the working directory:
+```bash
 agent-browser screenshot ".sparkecode-artifacts/screenshots/page.png"
 agent-browser screenshot ".sparkecode-artifacts/screenshots/fullpage.png" --fullpage
 ```
 
-After taking a screenshot, use `read_file` to view it:
+After taking a screenshot, view it with `read_file` and upload it with `upload_file` to get a shareable URL:
 ```
 read_file({ path: ".sparkecode-artifacts/screenshots/page.png" })
+upload_file({ path: ".sparkecode-artifacts/screenshots/page.png", name: "page.png" })
 ```
-The `read_file` tool supports images (png, jpg, gif, webp) and will display them visually so you can analyze the page.
+The `read_file` tool supports images (png, jpg, gif, webp) and will display them visually so you can analyze the page. The `upload_file` tool returns a `downloadUrl` — use this URL (not the local path) in task results and reports.
 
 ### Waiting
 ```bash

package/dist/tools/index.js

@@ -37,7 +37,8 @@ var init_types = __esm({
       status: z.enum(["running", "completed", "failed"]),
       result: z.unknown().optional(),
       error: z.string().optional(),
-      iterations: z.number().optional()
+      iterations: z.number().optional(),
+      parentTaskId: z.string().optional()
     });
     SessionConfigSchema = z.object({
       toolApprovals: z.record(z.string(), z.boolean()).optional(),