Botpress discloses breach that exposed bot configs in 16 customer workspaces
Company post: attacker abused an internal auth-proxy flaw and an employee API credential Oct 13–21; configs can hold third-party secrets; customers told to rotate credentials in affected bots.
Why it matters
Botpress says the attacker probed an authentication proxy from October 12, forged staff-looking sessions, then on October 21 used an employee API credential with admin rights to read configurations in 16 workspaces and change integration settings in two more. The company attributes motive to AI-credit resale, reports no ransom and no evidence of published customer data, and says source code and build pipelines were untouched. Containment began the same day; credential rotation and staff-access hardening continue. It is a concrete reminder that agent platforms store third-party keys inside bot configs.
Key facts
- Disclosure posted Nov 6
- Auth-proxy flaw exploited from Oct 12
- Employee API credential used Oct 21 07:44–08:23 UTC
- 16 workspaces configs read; 2 more integration changes
- Customers told to treat stored credentials as exposed