The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through search results retrieved from the agent registry. • Ingestion points: The scripts/search.js script fetches agent metadata (names, descriptions, tags) from an external search service. • Boundary markers: No explicit delimiters or guardrails are implemented to isolate the fetched data from the agent's logic. • Capability inventory: The skill has the capability to execute blockchain transactions via register.js, feedback.js, and set-uri.js using the user's private key. • Sanitization: Metadata is processed for display without semantic sanitization, potentially allowing malicious content in agent descriptions to influence the agent's behavior.
[EXTERNAL_DOWNLOADS]: The scripts/search.js script makes network calls to the vendor's search API at https://tn-search-service.bankofai.io/api/v1/search to retrieve agent registry information.
[DATA_EXFILTRATION]: The skill's utility module scripts/utils.js is designed to read sensitive blockchain private keys from the local file system path ~/.clawdbot/wallets/.deployer_pk. While this is a documented mechanism for signing on-chain transactions, it involves reading sensitive credentials that must be managed with care.

8004-skill