Observed arrival · 2026-09-26
PyxGrant puts a signed checkpoint in an AI agent’s path
PyxGrant presents a self-hosted control layer that checks coding-agent and MCP tool calls, then allows, redacts, holds, or blocks them.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams running coding agents with MCP tools
- Worth noticing
- The browser demo signs decisions with a key generated in the browser; some machine-level checks are unavailable there and fail closed.
Field notes
The proposed first deployment centers on coding agents and MCP servers found on developer laptops and CI runners. The page describes discovered servers staying off until approved, and a shared run budget that follows child processes carrying the same run ID. It also says some checks need access to the real machine; those are unavailable in the browser demo and are described as failing closed.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
◎NicheUnusually specific use
One card from the complete issue