The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automates a complex sequence of system commands for dependency management and testing, including just upgrade, mise upgrade, uv sync, and playwright install.
[CREDENTIALS_UNSAFE]: The skill retrieves a sensitive GitHub authentication token using the command mise token github --raw to facilitate automated updates to repository workflows and actions.
[EXTERNAL_DOWNLOADS]: The skill fetches tools and configuration files from external sources, such as the npm-check-updates utility via pnpm dlx and rule definitions from the iloveitaly/llm-ide-rules repository.
[INDIRECT_PROMPT_INJECTION]: The skill ingests instructions from an external repository, which are then used to influence agent behavior across multiple IDE clients.
Ingestion points: The llm-ide-rules download command fetches instructions.md and commands.md from a remote source into the local workspace.
Boundary markers: No explicit delimiters or safety warnings are provided to separate these external instructions from the agent's core guidelines.
Capability inventory: The skill possesses broad permissions, including shell execution, file system modification, and network access via package managers.
Sanitization: The skill does not perform validation or sanitization on the downloaded instruction files before they are applied to the agent's environment.

just-upgrade