@jskit-ai/create-app 0.1.78 → 0.1.80

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jskit-ai/create-app",
-  "version": "0.1.78",
+  "version": "0.1.80",
   "description": "Scaffold minimal JSKIT app shells.",
   "type": "module",
   "files": [

package/src/server/index.js

@@ -631,7 +631,7 @@ export async function runCli(
       stdout.write("Next steps:\n");
       stdout.write(`- cd ${shellQuote(targetLabel)}\n`);
       if (isAiSeedTemplate) {
-        stdout.write("- hand the repo to your AI agent and have it read AGENTS.md\n");
+        stdout.write("- use jskit session after the real scaffold exists\n");
         stdout.write(
           `- once the Stage 1 shape is clear, let it run npx @jskit-ai/create-app ${shellQuote(result.appName)} --target . --force --tenancy-mode <mode>\n`
         );

package/templates/ai-seed/AGENTS.md

@@ -1,39 +1,12 @@
 # AI Seed Instructions
 
-This workspace is intentionally only a JSKIT seed scaffold. It is not a real JSKIT app yet.
+This workspace is only a JSKIT seed scaffold. It is not a runnable JSKIT app yet.
 
-Use this file only for the initial JSKIT setup conversation.
+Keep this stage minimal:
 
-If you need more JSKIT documentation before the real scaffold exists, use:
+- Decide the app name and tenancy mode.
+- Promote the directory into a real scaffold with `npx @jskit-ai/create-app <app-name> --target . --force --tenancy-mode <mode>`.
+- Run `npm install`.
+- If local JSKIT checkout links are used, run `npm run devlinks` immediately afterward.
 
-- `https://github.com/mobily-enterprises/jskit-ai/blob/main/packages/agent-docs/site/guide/index.md`
-- `https://github.com/mobily-enterprises/jskit-ai/blob/main/packages/agent-docs/site/guide/app-setup/initial-scaffolding.md`
-- `https://github.com/mobily-enterprises/jskit-ai/blob/main/packages/agent-docs/site/guide/app-setup/quickstart.md`
-- `https://github.com/mobily-enterprises/jskit-ai/blob/main/packages/agent-docs/site/guide/app-setup/working-with-the-jskit-cli.md`
-
-What to do now:
-
-1. Ask only the Stage 1 setup questions first:
-   - app purpose
-   - tenancy mode: `none`, `personal`, or `workspaces`
-   - database engine: MySQL or Postgres
-   - auth provider
-   - whether the baseline should include users, console, workspaces, realtime, or assistant
-2. If a database-backed baseline is selected, do not stop at the engine name. Confirm one of these is true before moving on:
-   - the target database already exists and the app connection values are known
-   - or the developer has enough local admin or create-database access for the chosen engine to create it now
-3. Once the baseline package stack is clear, ask plainly for the exact local development setup values needed for the next chosen install step. Use the real env var names or option names instead of vague requests for "credentials".
-4. Treat those values as routine local development setup inputs, not production secrets. Examples:
-   - `DB_HOST`, `DB_PORT`, `DB_NAME`, `DB_USER`, `DB_PASSWORD`
-   - `AUTH_SUPABASE_URL`, `AUTH_SUPABASE_PUBLISHABLE_KEY`, `APP_PUBLIC_URL`
-   - `OPENAI_API_KEY` if assistant support is selected
-5. For MySQL or Postgres, if the database does not exist yet, ask for whatever local admin or create-database access is needed to create it, then make sure the database exists before promoting the scaffold. Do not move on while the database is still hypothetical.
-6. If assistant support is selected, explain that `OPENAI_API_KEY` can be given now or wired later; the scaffold can still be created first.
-7. Do not hand-write JSKIT runtime files or invent file topology in this seed workspace.
-8. Once the Stage 1 decisions are clear, promote this directory into the real app scaffold by running:
-   `npx @jskit-ai/create-app <app-name> --target . --force --tenancy-mode <mode>`
-   Use the current directory name as `<app-name>` unless the developer explicitly wants a different app name.
-9. After that, run `npm install`.
-10. If this app is being tested against a local JSKIT checkout with linked packages, run `npm run devlinks` immediately after `npm install`.
-11. Then stop, re-read the overwritten `AGENTS.md`, and continue with the normal JSKIT app workflow from that file. Do not rely on memory from the seed stage.
-12. Do not install JSKIT runtime packages until the new app `AGENTS.md` has been re-read.
+After the real scaffold exists, use `jskit session` for issue work. Do not continue from this seed file as a workflow source of truth.

package/templates/base-shell/AGENTS.md

@@ -1,11 +1,11 @@
-# App Agent Wrapper
+# App Agent Instructions
 
-Read and follow:
+Use JSKIT sessions for issue work.
 
-- `node_modules/@jskit-ai/agent-docs/templates/app/AGENTS.md`
+- `jskit session`
+- `jskit session create`
+- `jskit session <session_id>`
+- `jskit session <session_id> step`
+- `jskit session <session_id> step --json` for tools and UIs
 
-If that file does not exist yet:
-
-- install dependencies so `node_modules/@jskit-ai/agent-docs/` exists
-- inspect the workspace before assuming a JSKIT app already exists
-- if the workspace is empty and the user wants a new app, start with high-level initialize questions only
+Follow the current session step output and prompt. Do not invent a parallel manual issue workflow.

package/templates/base-shell/gitignore

@@ -2,7 +2,6 @@ node_modules/
 dist/
 coverage/
 test-results/
-.jskit/WORKBOARD.md
 .jskit/verification/
 .env
 .env.*