The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands via the Bash tool to interact with hoyeon-cli, a specialized vendor tool for project specification management. These commands are limited to state initialization (spec init), data merging (spec merge), and schema validation (spec validate).
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input during the 'Interview Loop' and incorporates this data into prompts for sub-agents, such as the L2-reviewer tasked with checking decision clarity.
Ingestion points: User goal inputs and answers to scenario-based questions collected via the AskUserQuestion tool.
Boundary markers: The skill employs a structured JSON schema (v1) to delimit user-provided content from instructions, though the sub-agent prompts use standard interpolation.
Capability inventory: The agent has access to Write, Bash, and Task tools across various scripts.
Sanitization: While no explicit string sanitization is mentioned, the skill requires mandatory human approval at three distinct gates (L2, L3, and L4) before the final plan can be executed, providing a robust manual oversight mechanism.

specify