The Agent Skills Directory

[DATA_EXFILTRATION]: The instructions direct the agent to "explore the repository" and "skim relevant source files" including "config" and "models". Accessing configuration files is a sensitive operation as they often contain environment variables or internal architectural details.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the local repository to ground the generated stories in reality. The lack of boundary markers or specific instructions to ignore embedded commands in these files creates a vulnerability to indirect prompt injection.
Ingestion points: Repository README, documentation folders, and source code files (routes, components, models, config).
Boundary markers: Absent; the skill does not define specific delimiters or instructions to ignore instructions found within the codebase.
Capability inventory: The skill utilizes file system read access and generates user-facing text output.
Sanitization: None; the content from the files is directly used to influence the generated output without filtering.
[EXTERNAL_DOWNLOADS]: The skill manifest includes an installation command that fetches the SKILL.md file from a GitHub repository (github.com/razbakov/skills).

user-story