gigaverse
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- Credentials Unsafe (MEDIUM): The skill generates and stores blockchain private keys in
~/.secrets/gigaverse-private-key.txt. While the scripts set restricted file permissions (chmod 600), the keys are stored in plain text on the local filesystem. - Command Execution (MEDIUM): In
scripts/auth.shandscripts/setup-wallet.sh, the private key is passed as a string literal to anode -ecommand. This makes the sensitive private key visible to any user or process on the system capable of viewing the process list (e.g., via thepscommand). - Data Exposure (LOW): The skill accesses the sensitive hidden directory
~/.secrets/to manage wallet credentials. - Indirect Prompt Injection (LOW): The skill processes data from external game APIs (gigaverse.io) to track game state and loot. This represents a potential attack surface where a compromised API could influence agent decision-making.
- Ingestion points: API responses from
https://gigaverse.io/apiprocessed inHEARTBEAT.mdandreferences/run-tracking.md. - Boundary markers: None identified.
- Capability inventory: Shell execution (
curl,node), signing blockchain messages, and sending ETH transactions viapurchase-juice.ts. - Sanitization: Relies on
jqfor structural parsing but does not explicitly sanitize string content from the API.
Audit Metadata