Observed arrival · 2026-11-18
Drawer Two tests whether agents can bypass spending guards
A local CLI and MCP tool runs adversarial scenarios against AI agents’ buy, sign, and budget guards.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams testing AI-agent spending and signing controls
- Worth noticing
- Runs can use synthetic scenarios or staging HTTP with a profile and environment variables supplied by the operator.
Field notes
The page distinguishes quick fixture runs from staging HTTP tests, which require an operator-supplied profile and environment variables. Its result scheme includes a NOT SURE state for missing observations rather than treating every run as a pass or failure. The homepage links to Linux and macOS binaries, checksums, source, and setup instructions for both the CLI and Cursor MCP route.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue