The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches agent reputation, Sybil risk, and historical data from the TrustLayer API (api.thetrustlayer.xyz). These are data-only requests used to inform the agent's decision-making process.
[COMMAND_EXECUTION]: The skill uses curl to interact with the service endpoints. The output is processed using standard utilities like grep, cut, and jq for field extraction, rather than being piped directly to a command interpreter.
[PROMPT_INJECTION]: The skill is designed to ingest and act upon data from an external API, creating a surface for indirect prompt injection.
Ingestion points: Data enters the agent context via curl requests to the TrustLayer API (e.g., /trust, /agent, /reviewer endpoints).
Boundary markers: None explicitly defined to separate external data from agent instructions.
Capability inventory: The skill uses curl to perform network operations but does not show evidence of file-writing or subprocess execution beyond data retrieval.
Sanitization: The skill uses targeted extraction (grep, cut) to isolate specific JSON fields (e.g., sybil_risk, trust_score), which limits the impact of arbitrary content in other fields.

trustlayer-sybil-scanner