The Agent Skills Directory

[SAFE]: The skill documents the use of internal platform primitives for state management, such as state::set, state::get, and state::update. These functions are used for typical application state persistence and do not involve unauthorized data access or exfiltration.- [PROMPT_INJECTION]: The skill facilitates the ingestion of data from an external state store via state::get and state::list (defined in SKILL.md). This constitutes an indirect prompt injection surface. 1. Ingestion points: state::get and state::list triggers. 2. Boundary markers: None specified in the instructions. 3. Capability inventory: The skill permits write operations including state::set, state::update, and state::delete. 4. Sanitization: No explicit sanitization or validation of retrieved data is described. This surface is considered a standard feature of state management and is assessed as safe in the absence of malicious instructions.

state-management