The Agent Skills Directory

[SAFE]: The skill uses official vendor resources, including the Tiramisu network explorer (tiramisu.explorer.arkiv.network) and the arkiv-graph package, which aligns with the author 'arkiv-network'.
[INDIRECT_PROMPT_INJECTION]: The skill processes entity data (payloads and labels) from the Arkiv network. It includes defensive instructions for developers to treat this data as untrusted and to 'render text safely' to prevent injection attacks during visualization.
Ingestion points: entities array in localGraph and result.entities in readSocialGraph (SKILL.md, references/fetch-and-refresh.md).
Boundary markers: The skill explicitly instructs the agent and developers to treat the processed data as untrusted.
Capability inventory: Visualization via React/d3 (react-force-graph-2d) and graph/table construction.
Sanitization: Instructions explicitly mandate safe text rendering for all untrusted data.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the arkiv-graph package via npm and provides a link to documentation on GitHub (SantiagoDevRel/arkiv-graph). These resources are documented neutrally as they align with the skill's stated purpose and the vendor's naming conventions.
[CREDENTIALS_UNSAFE]: The skill provides explicit security guidance to keep X-API-KEY server-side and uses placeholder hex strings for example addresses, ensuring no actual credentials or secrets are exposed.

arkiv-social-graph