The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data from release manifests and changelogs which could potentially contain malicious instructions.
Ingestion points: Accesses package.json, backstage.json, and remote content from versions.backstage.io and github.com.
Boundary markers: No explicit delimiters or instructions to ignore instructions within ingested files are provided to the agent.
Capability inventory: The skill executes commands through yarn and backstage-cli, including installation, build, and test operations.
Sanitization: The skill does not perform explicit sanitization of the retrieved text content before processing it.
[COMMAND_EXECUTION]: The skill uses backstage-cli and yarn to perform updates and verify code (e.g., versions:bump, versions:migrate, yarn tsc, yarn build, yarn test). These are standard development tasks for the targeted ecosystem.
[EXTERNAL_DOWNLOADS]: The skill fetches data from well-known services such as https://versions.backstage.io and https://github.com/backstage/backstage/releases. These are official resources for the Backstage project and are considered well-known services in this context.

rhdh-backstage-upgrade