The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill acts as an interface to external databases, creating a surface for indirect prompt injection.
Ingestion points: Database query results (e.g., from dbcli query) are returned to the agent's context.
Boundary markers: No specific delimiters are mandated for wrapping database output to distinguish it from instructions.
Capability inventory: The agent can execute powerful commands via dbcli, including data deletion and schema changes.
Sanitization: Content retrieved from databases is not sanitized before processing.
[EXTERNAL_DOWNLOADS]: The documentation references external sources for the required CLI binary and deployment scripts.
Evidence: Mentions of a GitHub repository (your-repo/dbcli) and deployment scripts (deploy-skills.ps1, deploy-skills.py) used to install the dbcli tool.
Context: These are standard references for the tool's installation and are not considered malicious.
[COMMAND_EXECUTION]: The skill's primary functionality is derived from executing the dbcli command-line tool.
Evidence: All core operations and sub-skills invoke the dbcli binary to interact with databases.
Context: This execution is the intended purpose of the skill and is properly scoped within the allowed-tools configuration.

dbcli