The Agent Skills Directory

[SAFE]: The skill uses a restricted set of tools (Read, Grep, Glob, Write, AskUserQuestion) to perform its task of goal generation.
[SAFE]: Documentation in references/design-notes.md demonstrates a security-conscious design by explicitly removing a requirement for a privileged sudoers-based helper binary that existed in the original Codex port.
[SAFE]: The skill limits its file writing operations to the .agents/goals/ directory and includes automated slug sanitization (lowercase ASCII, strict safe pattern [a-z0-9][a-z0-9._-]{0,80}) to prevent path traversal or malicious filename creation.
[SAFE]: Model invocation is disabled by default (disable-model-invocation: true), ensuring the skill only runs when explicitly requested by the user, preventing unauthorized context poisoning.

goalify