The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill provides instructions and scripts designed to automatically scan the project environment for sensitive configuration files, including .env, settings.py, and alembic.ini, specifically to extract database connection strings and credentials.
Evidence: This behavior is documented in SKILL.md under the 'Detect Database Configuration' section and in references/database-configs.md.
[DYNAMIC_EXECUTION]: The skill's configuration template utilizes string-encoded lambda functions for generating data, indicating that the orchestrator performs runtime evaluation (eval/exec) of these configuration strings.
Evidence: Found in assets/seed-config-template.yaml where fields like username are mapped to values such as lambda fake: fake.user_name().
[COMMAND_EXECUTION]: The seeding configuration schema includes a hooks section that allows for the definition and execution of arbitrary shell commands after the seeding process completes.
Evidence: assets/seed-config-template.yaml defines a hooks section with examples such as python scripts/update_search_index.py.
[INDIRECT_PROMPT_INJECTION]: The skill features a tool for automated database schema inspection that ingests untrusted metadata (table and column names) to generate Python factory code and JSON fixtures, creating a risk if the metadata contains malicious instructions.
Ingestion points: Database schema metadata ingested by scripts/inspect_schema.py (referenced in SKILL.md).
Boundary markers: None described for the schema analysis or code generation steps.
Capability inventory: The skill can write files (factory generation) and perform database writes (seeding).
Sanitization: No sanitization of table or column names derived from the external schema is described before they are used to generate code.

db-seeder