The Agent Skills Directory

[PROMPT_INJECTION] (HIGH): The skill is highly susceptible to Indirect Prompt Injection (Category 8) because it ingests untrusted user requirements to generate SQL code that is subsequently executed against a database via the Supabase CLI. An attacker could inject malicious SQL logic (e.g., exfiltration triggers or RLS bypasses) into the 'pattern requirements'. * Ingestion points: User-provided pattern requirements processed in SKILL.md. * Boundary markers: Absent; there are no delimiters or instructions to ignore embedded commands in the SQL templates. * Capability inventory: generate-schema.sh (file write), apply-migration.sh (executes supabase db push). * Sanitization: validate-schema.sh is present but only checks for stylistic best practices and naming conventions, not malicious SQL intent or logic.
[COMMAND_EXECUTION] (MEDIUM): The scripts generate-schema.sh and apply-migration.sh take file paths as command-line arguments and use them directly in shell operations (cat, cp, sed) without sanitization. An attacker could provide manipulated paths (e.g., path traversal or paths starting with hyphens) to overwrite unintended local files or inject flags into the sed command.

schema-patterns