arcium-program-development

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill explicitly supports using CircuitSource::OffChain with a user-supplied URL (see "Offchain Circuit Flow" in references/implementation-playbook.md and the offchain guidance in references/docs-and-migrations.md), which causes the system to fetch and consume external circuit artifacts from open URLs and therefore exposes the agent/runtime to untrusted third‑party content.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill allows fetching offchain circuit artifacts at runtime via CircuitSource::OffChain (e.g., init_comp_def(... CircuitSource::OffChain(OffChainCircuitSource { source: "<CIRCUIT_URL>".to_string(), hash: circuit_hash!("<IX_NAME>") }) ...)), which would load and execute remote circuit code as a required runtime dependency for offchain-circuit flows (e.g., encrypted_dna_matching).