The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines an interface for the agentic-db CLI tool. It instructs the agent to execute shell commands for context management, authentication via tokens, and database operations. The commands are designed for non-interactive use using the --tty false flag to ensure parseable JSON output.
[SAFE]: The skill documents standard database management functionality. It includes a surface for indirect prompt injection through search and retrieval commands (agentic-db search and agentic-db ask), which is inherent to its intended use case. Analysis of this surface reveals: 1. Ingestion points: Results from database queries and RAG-based answers (e.g., in SKILL.md and references/chunk.md). 2. Boundary markers: None specified in the instructions to delimit retrieved content. 3. Capability inventory: The agent has extensive permissions to create, update, and delete records across 91 tables including contacts, tasks, and system messages. 4. Sanitization: Not explicitly defined in the skill logic.

cli-default