The Agent Skills Directory

[PROMPT_INJECTION]: The skill's description field attempts to override agent behavior by stating the skill is "REQUIRED" and should be used "even if the user doesn't explicitly mention 'seed' or 'SQL'", which is a pattern designed to bypass standard task triage.
[DYNAMIC_EXECUTION]: The skill provides Python and Go code snippets in references/password-hashing.md and explicitly instructs the agent to execute them locally using commands like `python3
<<'PY'andgo run /tmp/genhash` to verify or generate password hashes.
[COMMAND_EXECUTION]: The instructions in references/password-hashing.md involve executing multiple shell commands, including printf, wc -c, and cat, to manipulate and verify hash strings.
[CREDENTIALS_UNSAFE]: The file references/password-hashing.md contains several hardcoded bcrypt hashes (e.g., $2a$10$0Ah2j61iId12kpgsbARXoevqxQrR0FLxJoQJJ9hDiPle0C2a6I.te) as pre-generated samples. While documented for development use with the password "Passw0rd!", they are static hardcoded secrets.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external project files which may contain untrusted data.
Ingestion points: The skill reads Ent schema files (ent/schema/*.go), migration scripts, and product documentation to build its model.
Boundary markers: There are no explicit instructions or delimiters defined to protect the agent from instructions embedded within these external files.
Capability inventory: The skill allows the agent to generate and execute local scripts and write SQL artifacts.
Sanitization: The skill mentions validating enums and JSON structure, but does not provide logic to sanitize the input files against prompt injection.

ent-seed-sql-generator