debridge
Pass
Audited by Gen Agent Trust Hub on Apr 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches the deBridge Solana SDK directly from the official deBridge Finance GitHub repository using the cargo dependency manager (
github.com/debridge-finance/debridge-solana-sdk.git). - [DATA_EXFILTRATION]: The skill exposes functions for sending arbitrary data buffers across blockchains (
send_message,invoke_debridge_send). While this is the intended purpose of the protocol, it enables the transmission of data from the agent's environment to external networks. - [PROMPT_INJECTION]: The skill implements cross-chain message processing, which creates an indirect prompt injection surface where the agent may ingest data containing malicious instructions from other blockchains.
- Ingestion points: The
receive_messageandreceive_governancefunctions inexamples/message-passing/README.mdare entry points for untrusted data from external chains. - Boundary markers: The provided code examples lack boundary markers or explicit instructions for the agent to ignore command-like content within received messages.
- Capability inventory: The skill enables cross-chain transfers and arbitrary contract calls (
invoke_debridge_send,invoke_init_external_call) which could be triggered by processed data. - Sanitization: No payload validation or instruction filtering is implemented in the provided message processing examples.
Audit Metadata