Observed arrival · 2026-11-09
toolwrap puts a human checkpoint between AI agents and MCP tools
A local MCP gateway for coding agents that separates tools and credentials by project and asks for approval before risky calls.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- People running coding agents across multiple projects
- Worth noticing
- A risky call is denied after four minutes without a response; edits that weaken protections also wait for approval.
Field notes
According to the page, each project uses a TOML set in ~/.mcphub/sets; toolwrap launches its listed MCP servers and exposes them to the agent as one MCP server. The examples show tool-level controls such as disabling delete_issue, pinning Jira searches to a project, and requiring confirmation for database queries. The site says credentials remain in the OS keychain, calls are logged with secrets redacted, and CLI controls mirror the desktop window.
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