The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions direct the agent to scan for connection strings in sensitive project files, specifically mentioning .env, database.yml, settings.py, and config/ directories to detect the database engine.
[COMMAND_EXECUTION]: The skill utilizes Bash, Write, and Edit tools to generate DDL scripts, ORM configurations, and migration files. This includes executing migration tools like Prisma, Drizzle, or Alembic to generate schema revisions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted user-provided domain descriptions and translates them into file-system changes and potentially shell commands.
Ingestion points: User-provided domain descriptions processed in Step 1 of the SKILL.md instructions.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the prompt templates.
Capability inventory: Use of Bash, Write, Edit, and Task tools to modify project files and potentially execute environment commands.
Sanitization: No sanitization, escaping, or validation of the domain description content is performed before interpolation.

flux-schema