The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs its primary function by interacting with the ThinkingData analytics platform through the ae-cli command-line tool. It executes shell commands to discover projects, retrieve event metadata, and run analytical queries. These operations are well-documented and restricted to the authenticated environment of the user's ThinkingData project.\n- [DYNAMIC_EXECUTION]: The skill uses a template-based approach to generate SQL queries at runtime. It populates templates from references/sql-templates.md with user-provided variables like drama names and date ranges. The skill specifically addresses security risks associated with this pattern by instructing the agent to escape all single quotes in user input to prevent SQL injection attacks within the CLI definition payloads.\n- [INDIRECT_PROMPT_INJECTION]: As the skill is designed to ingest and summarize external data from ThinkingData events, it possesses a standard attack surface for indirect prompt injection.\n
Ingestion points: Data enters the agent context through the outputs of ae-cli analytical commands executed during the analysis workflow.\n
Boundary markers: The skill uses structured JSON command outputs and predefined report templates to maintain data separation.\n
Capability inventory: The skill's capabilities include executing authenticated shell commands via ae-cli and managing temporary query definition files.\n
Sanitization: While the skill enforces sanitization for inputs used in SQL construction, it does not explicitly specify sanitization routines for the data returned from the ThinkingData API before it is processed by the agent.

drama-retention-analyzer