myte 0.0.15 → 0.0.17

package/README.md

@@ -88,6 +88,8 @@ Run this first when onboarding on a new machine:
 
 Use `myte ai` as the thin inference gateway client. The npm package does not ship models or backend internals; it sends your request to `api.myte.ai` with `MYTEAI_API_KEY`.
 
+`--json-response` asks for strict JSON and now sends both `myte_json_response: true` and the OpenAI-compatible `response_format: { type: "json_object" }` shape. On the public developer API, that strict-JSON path also defaults to `medium` reasoning unless the caller explicitly overrides reasoning.
+
 Examples:
 
 - `npx myte ai "Explain what this repository does"`
@@ -189,13 +191,15 @@ What it does:
 - fetches a project-scoped feedback snapshot from the API using your project key
 - defaults to open `Pending` feedback
 - writes one deterministic feedback snapshot under `MyteCommandCenter/data/feedback.yml`
-- includes readable PRD markdown inline when available
+- keeps feedback metadata plus conversation turns in `feedback.yml`
+- writes full PRD markdown/text into `MyteCommandCenter/PRD/feedback-sync/*.md` and stores a pointer in `feedback.yml`
 - fully replaces the aggregated feedback sync file so stale feedback does not accumulate locally
 - does not delete `MyteCommandCenter/data/missions/*.yml`
 
 Files written:
 - `MyteCommandCenter/data/project.yml`
 - `MyteCommandCenter/data/feedback.yml`
+- `MyteCommandCenter/PRD/feedback-sync/*.md`
 
 Useful forms:
 - `npx myte feedback-sync --json`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "myte",
-  "version": "0.0.15",
+  "version": "0.0.17",
   "description": "Myte developer CLI (Project Assistant + Myte AI gateway).",
   "type": "commonjs",
   "license": "MIT",
@@ -16,6 +16,6 @@
     "node": ">=18"
   },
   "dependencies": {
-    "@mytegroupinc/myte-core": "0.0.15"
+    "@mytegroupinc/myte-core": "0.0.17"
   }
 }