The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a process for ingesting user-provided descriptions and transforming them into structured graph nodes (e.g., Kriya, Phenomenon, Vimarsha). This creates a data ingestion surface; however, the skill implements a 'Gate of Novelty' and rigorous ontological verification steps that require the agent to justify the necessity and purpose (telos) of the data before using tool calls like iskron_add_kriya. These logic gates serve as effective boundaries against the mechanical replication of potentially malicious content within the graph structure.
[DATA_EXPOSURE]: The skill instructions mention a realm=<token> requirement for tool invocations. It directs the agent to confirm the target graph ('realm') before performing any actions, which is a standard security practice for scoping operations and preventing accidental data leakage across different environments. No hardcoded credentials or sensitive file paths were detected.
[PROMPT_INJECTION]: The skill uses natural language triggers (e.g., 'запиши', 'create phenomenon') to initiate its workflow. These triggers are used to activate legitimate recording functions and do not contain instructions to override system safety guidelines or bypass agent constraints.

writing