delphi

Pass

Audited by Gen Agent Trust Hub on Apr 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches blockchain state, market pricing, and historical trade information from official Gensyn API endpoints, the Goldsky subgraph, and Alchemy RPC nodes. These are standard, well-known services required for the skill's primary functionality.
  • [COMMAND_EXECUTION]: Provides a suite of TypeScript scripts for executing on-chain transactions (buys, sells, redemptions, liquidations), asset bridging, and balance checks. These scripts are invoked via npx tsx and require the user to have previously configured a wallet with signing capabilities.
  • [PROMPT_INJECTION]: The skill processes untrusted market metadata which presents an indirect prompt injection surface.
  • Ingestion points: Market metadata (questions, titles, outcome labels) is fetched in scripts/list-markets.ts, scripts/get-market.ts, and scripts/list-positions.ts.
  • Boundary markers: Absent; metadata is interpolated directly into the agent's context without explicit delimiters or instructions to ignore embedded commands.
  • Capability inventory: The agent can execute transactions that move funds (buys/sells) and modify contract state via the provided scripts.
  • Sanitization: No explicit sanitization of fetched string data is observed in the implementation scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 23, 2026, 01:36 PM