The Agent Skills Directory

[SAFE]: The skill consists of instructional documentation authored by TencentCloudBase for its own database services. It contains no malicious code, obfuscation, or exfiltration patterns.
[NO_CODE]: The skill does not include any executable scripts, binaries, or configuration files; it is composed entirely of markdown instructions for agent behavior.
[COMMAND_EXECUTION]: The skill instructs agents on how to execute SQL commands via executeWriteSQL. Security risks are mitigated by the inclusion of safety protocols requiring agents to summarize destructive actions and perform preliminary data validation using executeReadOnlySQL.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow where agents process results from database queries (executeReadOnlySQL), creating an attack surface for indirect prompt injection if the database contains adversarial content.
Ingestion points: Results from executeReadOnlySQL are processed by the agent.
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands in query results.
Capability inventory: The agent has the ability to modify the database (executeWriteSQL) and change table permissions (writeSecurityRule).
Sanitization: No explicit sanitization or filtering instructions for query results are provided in the skill body.

relational-database-mcp-cloudbase