The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by design, as it ingests and processes untrusted data from multiple sources to drive system modifications.
Ingestion points: The skill processes user-provided "prosaic artifacts" (Mode 3: reports, roadmaps, telos) and external graph data retrieved via tools like iskron_orient and iskron_look (SKILL.md).
Boundary markers: There are no explicit instructions or delimiters defined to separate untrusted external content from the agent's core instructions, nor are there warnings for the agent to ignore embedded commands within the analyzed artifacts.
Capability inventory: The skill possesses the ability to modify the permanent records of the system through write-capable tools such as iskron_add_vimarsha (creating new entries), iskron_update (modifying existing nodes), and iskron_arrow (creating relationship links).
Sanitization: The instructions do not specify any validation, sanitization, or filtering of the external content before it is processed or used to populate descriptions in written records.

integrity