mantle-portfolio-analyst
Pass
Audited by Gen Agent Trust Hub on Nov 8, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
mantle-clicommands to fetch account balances, token holdings, and allowances. These operations are strictly limited to read-only functions, and the instructions explicitly forbid transaction construction or network connectivity beyond the intended ledger queries. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data in the form of wallet addresses and token/spender lists. It mitigates injection risks through mandatory validation steps:
- Ingestion points: Wallet addresses and token/spender scopes provided via user input (documented in
SKILL.mdandreferences/rpc-readonly-workflow.md). - Boundary markers: Output is processed as structured JSON (
--jsonflag) to prevent parsing confusion. - Capability inventory: Capabilities are limited to read-only CLI calls for blockchain state (
balance,token-balances,allowances). - Sanitization: The skill implements checksum validation for addresses and uses
mantle-cli registry validateto verify input formats before processing. - [SAFE]: The workflow incorporates security best practices, such as classifying approval risks with specific heuristic rules and requiring explicit rationale for risk levels. It correctly identifies high-risk unlimited approvals (value >= 2^255) as a security signal for the user rather than a vulnerability in the skill itself.
Audit Metadata