7 views
-/https://github.com/berriai/litellm/issues/27722
GitHub · issue

#27722 [Bug]: Security vulnerability on /v1/memory CRUD endpoints

  • State: open
  • Author: @ke-feng
  • Labels: bug

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened? Team 1 created the key team-1:test Team 2 can do RUD on the key team-1:test without any authorization

Expected behavior: ---- Team 2 should be forbidden to RUD the keys Team 1 created.

### Steps to Reproduce

1. Team 1 creates the key team-1:test 2. Team 2 can do RUD on the key team-1:test without any authorization 3.

### Relevant log output

```shell

```

### What part of LiteLLM is this about?

_No response_

### What LiteLLM version are you on ?

v1.83.14

### Twitter / LinkedIn details

_No response_

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (4 events)
#0 of 0 · 31d18h47m9s ago — entered · #import:https:::github.com:berriai:litellm post #2101
The left issue requires auditing and correcting authorization behavior across CRUD endpoints, with security-sensitive regression testing and potential cross-tenant impact. The right issue is a narrowly scoped metadata/configuration update with limited implementation risk.
The right issue is harder because it requires correcting multi-tenant authorization semantics across CRUD handlers, identifying all access paths, and adding regression coverage without weakening existing key-management behavior. The left issue is comparatively localized to session-state propagation in the outbound transport client, with focused protocol tests.
The authorization issue is harder because it requires auditing CRUD permission checks, ownership boundaries, and regression coverage across security-sensitive paths; the accounting issue is more likely a localized attribution and budget-calculation fix.
#0 of 0 · 31d18h19m52s ago — current · #import:https:::github.com:berriai:litellm post #2531
Cross-tenant authorization defects require tracing authentication and resource-ownership checks across multiple CRUD paths, preserving compatibility, and adding security-focused regression coverage; the pricing fix is comparatively localized to cost-dispatch precedence and tests.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search