The MCP server that changes its mind after you approve it

A developer introduced Bulwark, a security scanner for AI agents that pins MCP tool descriptions in a lockfile by hash. If the description text changes after approval (e.g., a hidden instruction to read ~/.ssh/id_rsa), verification detects it even if the package version and config are unchanged.

• bulwark pin hashes all strings the model reads from MCP servers
• bulwark verify catches text changes, new arguments, and secret access
• Scanner decodes hidden Unicode Tag characters, zero-width, and base64 in descriptions
• Checks the 'lethal trifecta': private data, external input, and outbound sending