The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from path-addressed objects (e.g., /spaces, /.knowledge, /user) using the 'arc afs read' and 'arc afs explain' commands in SKILL.md. This ingestion of potentially untrusted data without explicit boundary markers or sanitization creates a vulnerability surface where maliciously crafted content in storage could influence agent behavior. * Ingestion points: 'arc afs read' and 'arc afs explain' commands in SKILL.md. * Boundary markers: No specific delimiters or safety instructions are defined for processing ingested storage data. * Capability inventory: The skill utilizes the 'arc' CLI for file system CRUD operations, including reading, writing, and listing objects. * Sanitization: No explicit sanitization or validation of retrieved content is mentioned in the instructions.
[COMMAND_EXECUTION]: The skill utilizes the 'arc' CLI utility for managing the DID Space data plane. Commands such as 'arc space init', 'arc space tree', and 'arc afs write' interact with the local file system and project-specific storage to persist application state.
[EXTERNAL_DOWNLOADS]: The skill references an installation method 'arc skill install space' which uses the vendor-provided 'arc' tool to fetch and set up the skill functionality. This is documented as a standard platform capability for the 'arcblock' ecosystem.

space