Mercury Blog · Author

tashiriskhan

Contributor, Mercury. Writes about agent permissions, security and running Mercury safely.

Sort

EngineeringOct 20, 20264 min read

The new bottleneck in AI coding isn't code generation. It's verification.

A Harvard working paper finds coding agents raised code output while review time rose 49%. Why verification, not generation, should decide whether an agent is useful.

tashiriskhan

SecurityOct 17, 20267 min read

Agents exceeded their permissions: how to scope them

CERT-In says AI agents read live business records at Indian firms. Here is how least privilege, scoped access, approvals and audit logs should limit them.

tashiriskhan