The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted repository data, including SQL migrations, JSON journals, and TypeScript configuration files. This creates a surface for indirect prompt injection where malicious instructions hidden in these files could attempt to influence the agent's decision-making process. The skill mitigates this through established safety protocols.
Ingestion points: The agent reads _journal.json, snapshot.json, SQL files, and drizzle.config.* files from the target repository.
Boundary markers: The skill instructions mandate a read-only "Diagnose" mode by default and require explicit confirmation for all destructive operations or automated repairs.
Capability inventory: The skill leverages git, python3 (for structural analysis), and rg (ripgrep).
Sanitization: The scripts/check_drizzle_migrations.py helper script performs structural parsing and regex-based content matching without executing or evaluating the file content.
[COMMAND_EXECUTION]: The skill executes a local Python helper script (scripts/check_drizzle_migrations.py) to perform structural integrity checks on migration directories. This script is designed with security best practices, including path traversal protection (using relative_to to restrict file access to the workspace) and a read-only focus that avoids database connections or code imports.
[DATA_EXFILTRATION]: The skill instructions include explicit requirements for the agent to identify and redact sensitive information. The agent is directed never to include database URLs, passwords, or tokens in its reports, instead using placeholders like <redacted> or describing the target environment in neutral terms. The helper script contains no network-capable code.

drizzle-migration-conflict