E2B
SDK / ToolOpen-source secure sandboxes where AI agents run code — the de facto execution layer for agent apps.
- sandbox
- code-execution
- sdk
- agents
"The AI proposes. The bytes decide." — a binary-analysis anti-hallucination verification layer, callable as an MCP server.
Reverify addresses a real failure mode: LLMs confidently invent binary-analysis facts. Its deterministic Python core (PE/ELF/Mach-O parsing, x86/x64/ARM disassembly, emulated execution) verifies the model's structural claims against the actual bytes; optional engines (capstone/unicorn/lief/Z3/angr) upgrade coverage. On a 19-file Windows system-binary benchmark the authors report AI textbook answers were 100% wrong and reverify caught every one with zero false positives. Ships as an MCP server for Claude Code/Cursor plus a CLI.
Repository
2akouwu/reverifyCharted
Open-source secure sandboxes where AI agents run code — the de facto execution layer for agent apps.
Pythonic framework for building MCP servers — decorators, type hints, and zero boilerplate.
Anthropic's official TypeScript SDK for building agentic applications.