switchboard

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill's templates and examples (e.g., examples/setup/example.ts, templates/setup.ts, and examples/feeds/pull-feed.ts) are configured to access the default Solana wallet private key file at ~/.config/solana/id.json. While this is standard practice for Solana development tools, it involves reading sensitive credentials from the filesystem.
  • [PROMPT_INJECTION]: The skill contains a minor metadata discrepancy; the creator field in the SKILL.md frontmatter is set to raunit-dev, which differs from the provided author context of JustaName-id.
  • [PROMPT_INJECTION]: The skill exhibits an Indirect Prompt Injection surface (Category 8) as it processes data from the external Switchboard oracle network.
  • Ingestion points: Data enters the agent's context through the CrossbarClient (for fetching signed oracle responses) and the SwitchboardSurge WebSocket client (for real-time price streaming) in examples/surge/streaming.ts.
  • Boundary markers: The instructions do not define explicit delimiters or 'ignore' instructions to wrap the external data ingested from the oracles.
  • Capability inventory: The skill's primary capability is interacting with the Solana blockchain via @coral-xyz/anchor and @switchboard-xyz/on-demand to build, sign, and send transactions.
  • Sanitization: While the Switchboard SDK performs cryptographic signature verification on the data, the skill does not implement additional prompt-level sanitization or validation for the values returned by the oracles.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 12:31 AM
Security Audit — agent-trust-hub — switchboard