Karchi Guard: an open-source check for what an AI agent can reach in a Supabase repo #52268
karthikkushi
started this conversation in
Show and tell
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
I built a small open-source CLI for a problem I kept seeing in AI-built Supabase apps: the agent working in the repo has far more power than anyone intended.
It reads your repo (and optionally
~/.claude.jsonand~/.cursor/mcp.json) and reports:service_role/sb_secret_keys an agent can read, or that ship to the browsersbp_access tokens and DB URLs with passwordsread_only=trueorproject_refsupabase db reset --linkedorsupabase projects deleteIt is read-only: no network, no database, and it never prints a secret. It does not check RLS. For that I point people to
supabase db advisors --type securityand rlsgrid.Repo: https://github.com/karthikkushi/karchi-guard
I would like feedback on the checks. Are there agent setups I am missing?
(I built this with help from Claude.)
All reactions