The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs the agent to request a Database Connection String (DSN) from the user and store it in a file named .dbmap-auto. DSNs typically contain sensitive information including hostnames, usernames, and passwords. Storing these in plaintext in a project-level file is an insecure credential management practice.
Evidence: printf '%s\n' '<confirmed_dsn>' > .dbmap-auto in SKILL.md.
[COMMAND_EXECUTION]: The skill executes a vendor-provided script to detect or list database connections. It relies on the environment variable REPOMAP_HOME or a default path in the user's home directory.
Evidence: "${REPOMAP_HOME:-$HOME/claude-repomap-command}/scripts/run.sh" dbmap --list in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill establishes a PostToolUse hook that automatically triggers when the agent runs database migration commands. This creates an automated processing pipeline where the output of third-party migration tools (e.g., Rails, Django, Prisma) is ingested by the agent to regenerate schema maps, representing a potential surface for indirect prompt injection if a migration tool's output is compromised.
Ingestion points: Output of shell migration commands (db:migrate, manage.py migrate, etc.).
Boundary markers: Not specified in the skill instructions.
Capability inventory: Shell command execution via vendor scripts, file write (DBMAP.md, .dbmap-auto).
Sanitization: No explicit sanitization of the migration tool output is mentioned before processing.

dbmap-auto-on