prism-wasm-integration

Pass

Audited by Gen Agent Trust Hub on Aug 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The integration guides include instructions for using package managers and build tools, such as pip install, npm install, and moon build, to set up the developer environment.\n- [EXTERNAL_DOWNLOADS]: The skill references external libraries and repositories hosted on GitHub and public package registries, which are intended to interface with the core logic.\n- [REMOTE_CODE_EXECUTION]: Code snippets in the documentation demonstrate how to load and instantiate WebAssembly modules, including patterns for fetching them from remote URLs in the TypeScript implementation.\n- [PROMPT_INJECTION]: The skill processes structured data from various LLM providers, creating a surface for indirect prompt injection where malicious instructions could be embedded in provider-specific JSON payloads.\n
  • Ingestion points: Functions like wasm_to_lux_req and wasm_lux_resp_to_provider in SKILL.md that ingest provider JSON.\n
  • Boundary markers: The Envelope contract defined in SKILL.md provides a consistent structure for return values and diagnostics to manage data flow.\n
  • Capability inventory: The skill enables loading and running a binary WASM module within the host application's process using runtimes such as wasmtime or wazero.\n
  • Sanitization: Implementation of diagnostic reporting and error checks within the suggested wrapper code to validate conversion fidelity.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 25, 2026, 02:22 PM
Security Audit — agent-trust-hub — prism-wasm-integration