@jelou/cli 1.42.3 → 1.44.0

package/README.md

@@ -11,11 +11,32 @@ Code, Cursor, Codex, and Windsurf on `jelou login`.
 
 ## Install
 
+**Requirements:** [Node.js](https://nodejs.org/) 18+ and npm. For the prompt
+route, an AI editor —
+[Claude Code](https://docs.anthropic.com/en/docs/claude-code), Cursor, Codex, or
+Windsurf.
+
+### Install with a prompt
+
+Open Claude Code, Cursor, Codex, or Windsurf and paste this. The agent does the
+rest:
+
+> Install the Jelou CLI: run **`npm install -g @jelou/cli`**, verify it with
+> `jelou --version`, then ask me for my Jelou API key (from
+> https://apps.jelou.ai/settings/api-keys) and run `jelou login --token <key>`
+> to authenticate. Finally run `jelou agent install` so the Jelou agent skills
+> are installed into this editor — then you can build, test, and ship Jelou
+> workflows for me from here.
+
+### Install manually
+
 ```bash
-npm install -g @jelou/cli
+npm install -g @jelou/cli   # requires Node.js 18+
+jelou login                 # paste your API key; skills auto-install
 ```
 
-Verify with `jelou --version`.
+Verify with `jelou --version`. Prefer to keep your key out of the chat history?
+The manual route reads it from a masked prompt.
 
 ## Sign up
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jelou/cli",
-  "version": "1.42.3",
+  "version": "1.44.0",
   "description": "Build AI agents on the Jelou platform — from your terminal or your AI editor.",
   "author": "Jelou Inc. (https://jelou.ai)",
   "license": "SEE LICENSE IN LICENSE",
@@ -23,11 +23,11 @@
     "THIRD_PARTY_NOTICES.md"
   ],
   "optionalDependencies": {
-    "@jelou/cli-darwin-arm64": "1.42.3",
-    "@jelou/cli-darwin-x64": "1.42.3",
-    "@jelou/cli-linux-x64": "1.42.3",
-    "@jelou/cli-linux-arm64": "1.42.3",
-    "@jelou/cli-win32-x64": "1.42.3"
+    "@jelou/cli-darwin-arm64": "1.44.0",
+    "@jelou/cli-darwin-x64": "1.44.0",
+    "@jelou/cli-linux-x64": "1.44.0",
+    "@jelou/cli-linux-arm64": "1.44.0",
+    "@jelou/cli-win32-x64": "1.44.0"
   },
   "keywords": [
     "jelou",
@@ -56,19 +56,27 @@
     "access": "public"
   },
   "jelou": {
-    "minVersion": "1.42.0",
+    "minVersion": "1.44.0",
     "release_notes": {
-      "version": "1.42.3",
-      "date": "2026-05-29",
-      "headline": "Human-handoff (CONNECT) nodes now route correctly; `jelou push` stops churning locked node titles.",
+      "version": "1.44.0",
+      "date": "2026-05-31",
+      "headline": "The local test dashboard is now interactive — start a chat with any workflow and watch it run live.",
       "highlights": [
         {
-          "type": "fixed",
-          "text": "Human-handoff (`CONNECT`) nodes built with the workflow skills now route to the assigned team or operator. Before, a handoff configured with the shorthand assignment passed validation but silently failed at runtime, so no agent was ever paged."
+          "type": "added",
+          "text": "`jelou test dashboard` can now start manual chats: open New chat, pick a workflow, and send messages — the bot's replies and the node trace stream in live, beside the runs your agent recorded. Runs recorded by the CLI stay read-only; chats you start in the dashboard are interactive."
         },
         {
-          "type": "fixed",
-          "text": "`jelou push` no longer rewrites locked node titles (like `TOOL` nodes) on every push, so repeated pushes stop showing spurious title changes."
+          "type": "improved",
+          "text": "The dashboard run view gained tester controls: a Messages/Debug toggle, an errors-only filter, a click-a-node payload drawer, and a spotlight on the first failed node so a broken run reads at a glance."
+        },
+        {
+          "type": "improved",
+          "text": "`jelou test send text`, `stats`, `chats`, and `trace` now point you to the dashboard so it's discoverable instead of hidden, and `jelou test dashboard --no-open` works as its help promised."
+        },
+        {
+          "type": "heads_up",
+          "text": "This release raises the minimum-version floor to 1.44.0. Any CLI below it shows a soft \"Update available\" prompt on the next run (and `meta.update_available.below_floor: true` for `--agent` callers). Nothing is blocked."
         }
       ]
     }