create-app_skill

Overview

This skill builds and deploys an Atris app from a natural language description. It guides creation of the app container, optional agent and skill, secrets storage, scheduling, and testing so you get a running chat, workflow, or autonomous app quickly.

How this skill works

I translate a user request into an Atris app specification: name, slug, UI template, instructions, and access mode. If the app needs external APIs or autonomous behavior, I store secrets, create or pick an agent, upload a markdown skill describing data access and processing, add the agent as a member, set a schedule, and run a verification test.

When to use it

• You want a chat widget or conversational app (screening, support).
• You need scheduled analytics or reports (daily Mixpanel insights).
• You want an autonomous workflow reacting to webhooks or data ingestion.
• You need a headless API-driven app to process CSVs or other batch inputs.
• You need a repeatable deployment process that includes secrets and monitoring.

Best practices

• Choose a clear app name and generate a hyphenated slug for the share_token.
• Set access_mode to private for internal workflows, public for shared chat apps.
• Store only required API keys and never display secret values after saving.
• Write the agent skill as a markdown file specifying data sources, analysis steps, outputs, and storage collections.
• Test with a manual trigger and inspect runs/status before enabling schedules or sharing.

Example use cases

• Daily Mixpanel analytics: private app, Mixpanel key, agent skill, daily cron, email summary.
• Chat-based candidate screening: public chat app, system instructions define interview flow, no agent required.
• Webhook-driven feedback processor: private app, agent analyzes ingested posts on trigger, outputs stored results.
• CSV lead qualification: upload via /ingest/batch, agent runs qualification skill on manual trigger, stores leads in a collection.