The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill's documentation recommends using sudo apt install during the setup process, which grants the agent or user administrative access to the host system for package management.
[EXTERNAL_DOWNLOADS]: The skill instructions involve downloading and installing the seekdb-cli package from the Python Package Index (PyPI), which is a standard public registry.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data retrieved from external database sources, creating a potential vulnerability surface.
Ingestion points: External data enters the agent's context through SQL query results, schema descriptions, and vector collection searches.
Boundary markers: Although the tool outputs structured JSON to help distinguish data from instructions, there are no explicit boundary markers or instructions to the agent to ignore embedded commands within the processed data.
Capability inventory: The skill possesses write capabilities to the database (sql --write), the ability to add new data from files or standard input (add), and the capacity to export database contents to the local file system (export).
Sanitization: The CLI tool provides built-in safety features, including automatic masking of sensitive fields (such as passwords and phone numbers) and a requirement for explicit flags to perform write or delete operations.
[COMMAND_EXECUTION]: The primary function of the skill is the execution of shell commands to interact with database engines, manage AI model endpoints, and perform data profiling.

seekdb-cli