The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to execute local utility scripts such as scripts/goal-design-packet.py and scripts/check-goal-design-packet.sh. These scripts handle packet creation, digest refreshing, and schema validation. The reliance on local scripts is a core part of the intended workflow and does not involve downloading or executing code from external sources.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it ingests human-provided objectives and processes them into agent-consumable packets.
Ingestion points: User-supplied <goal> input provided via the /goal-design trigger command in SKILL.md.
Boundary markers: The workflow uses double-quote wrapping when passing the goal to the creation script (scripts/goal-design-packet.py new --objective "<goal>"), but does not define explicit delimiters for the content inside the generated Markdown files.
Capability inventory: The skill is authorized to use Bash, Read, Grep, and Glob tools to manage files and execute validation logic.
Sanitization: The skill mitigates risks through a mandatory 'deterministic checker' that scans for schema violations and 'self-grading language.' It also requires an independent validation verdict before the packet can be consumed by downstream agents (/discovery or /plan).

goal-design