The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automates the application lifecycle using olares-cli, docker, git, and gh (GitHub CLI). These operations are core to its purpose of scaffolding, building, and deploying containerized applications.
[EXTERNAL_DOWNLOADS]: Downloads configuration and trusted container images from the beclab organization (e.g., beclab/terminal, beclab/docker, beclab/aboveos-busybox). These are authoritative resources from the skill's authoring vendor.
[PROMPT_INJECTION]: Features robust safety instructions ('Agent boundaries') in files like olares-chart-paid-apps.md and olares-chart-market-submit.md. These instructions explicitly forbid the agent from handling high-privilege secrets or executing on-chain transactions without the user's direct oversight.
[DATA_EXFILTRATION]: Performs legitimate network operations to well-known services (Docker Hub, GitHub Container Registry) and the Olares platform backend for the purpose of pushing images and uploading application charts. No unauthorized or suspicious data transmission was detected.
[CREDENTIALS_UNSAFE]: Appropriately handles platform secrets (like OLARES_USER_OPENAI_APIKEY or HF_TOKEN) by referencing them through the platform's environment variable system (valueFrom) rather than inlining or hardcoding values.

olares-chart