The Agent Skills Directory

Indirect Prompt Injection (LOW): The skill ingests untrusted data from local JSON/JSONL files (e.g., current.json, history/*.jsonl). Fields such as hostname, friendly_name, or notes in the scan output could be influenced by an attacker on the network to contain prompt injection payloads.
Ingestion points: current.json, history/*.jsonl, speedtests.jsonl via jq, rg, and awk commands.
Boundary markers: None. The instructions do not specify using delimiters or warnings to treat the data as untrusted.
Capability inventory: The skill uses subprocess execution for jq, rg, awk, and gzip. It does not perform network operations or file writes.
Sanitization: No sanitization or validation of the JSON string content is performed before the agent processes the tool output.

homescout-ingest