The Agent Skills Directory

[COMMAND_EXECUTION]: The skill identifies and executes shell commands defined in the target repository's configuration, such as linting and type-checking scripts.
[INDIRECT_PROMPT_INJECTION]: The skill is instructed to read and follow instructions found within the repository it is analyzing (Section 1.1). This introduces a risk where a malicious repository could provide instructions to override intended behavior.
Ingestion points: Repository-specific documentation and instruction files (referenced in SKILL.md).
Boundary markers: No explicit markers or delimiters are defined to separate repository content from core instructions.
Capability inventory: File system read/write access and shell command execution via bun and bunx (referenced in SKILL.md).
Sanitization: There is no explicit process mentioned for validating or sanitizing the instructions found within the repository.
[DYNAMIC_EXECUTION]: The skill executes arbitrary scripts from the repository's package.json file using bun run, which involves running code determined by the repository's contents.
[EXTERNAL_DOWNLOADS]: The skill utilizes bunx to fetch and run react-doctor from the npm registry if it is not already available locally.

fix-repository-diagnostics