The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent on how to retrieve and set blob attributes using the walrus get-blob-attribute and walrus set-blob-attribute commands. This creates an indirect prompt injection surface where the agent might ingest and act upon malicious instructions embedded in blob metadata by external actors.
Ingestion points: Untrusted metadata (attributes) is read from the Sui blockchain via the walrus CLI or SDK.
Boundary markers: The skill does not provide specific instructions to use delimiters or ignore instructions found within attribute values.
Capability inventory: The skill includes powerful capabilities such as walrus delete, walrus extend, and walrus burn-blobs which could be target actions for an injection attack.
Sanitization: No explicit sanitization or validation of the attribute keys or values is described.
[EXTERNAL_DOWNLOADS]: The skill references and fetches information from official Walrus project domains, including docs.wal.app for documentation and costcalculator.wal.app for storage estimation. These are official resources provided by the vendor, Mysten Labs.

walrus-blob-lifecycle