/

CVE-2026-92801

Severity:

HIGH RISK

(8.8)

Authorization Bypass (CWE-863) in cc-connect Feishu

CVE-2026-92801

Deferred

Vulnerability type

Category:

Authorization Bypass

CWE:

CWE-863

Impact:

Unauthorized agent command execution

Severity:

HIGH RISK

(8.8)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

cc-connect through version 1.5.0 does not apply per-user allowlist checks when processing Feishu interactive card callbacks. An attacker who can trigger a card action in an admitted chat may dispatch agent commands despite controls that protect the text-message handler.

Affected context

The issue occurs in deployments using cc-connect's Feishu integration when interactive card actions are accepted from chats that are admitted at the chat level. The onCardAction callback path may lack the per-user allowlist enforcement applied to ordinary text messages, allowing authorization decisions to differ between message types.

Exploit preconditions

  • Attack surface: Feishu interactive card callback handling
  • Required access: Ability to participate in or trigger actions within an admitted chat
  • Trigger point: The onCardAction handler processes an interactive card callback
  • Why exploitation is possible: The callback path does not consistently enforce the per-user allowlist before dispatching agent commands

Why this happens

The vulnerability results from inconsistent authorization enforcement across equivalent input paths. The text-message handler applies per-user allowlist filtering, while the interactive card callback handler can dispatch commands based on chat admission without applying the same user-level decision. This represents CWE-863, Incorrect Authorization, because a protected operation is reachable without validating the caller's permission in every relevant execution path.

Potential impact

  • An unauthorized user may dispatch agent commands
  • Attackers could cause actions to be performed under the agent's privileges
  • Sensitive information may be exposed through unauthorized commands
  • Connected systems or workflows could be modified
  • Abuse could affect other users or resources accessible to the agent

General mitigation guidance

Mitigation generally focuses on applying authorization consistently at the

point where agent commands are dispatched.

Common defensive patterns include:

  • Reusing a single centralized authorization check for text and card-action paths
  • Enforcing per-user allowlists after resolving the callback identity and before command dispatch
  • Treating chat admission as insufficient without user-level authorization
  • Applying least privilege to the agent and its integrations
  • Denying access by default when callback identity or authorization state is missing
  • Recording authorization decisions and rejected callback attempts for review

How teams detect issues like this

Detection approaches commonly include:

  • Static analysis and code review checks for command dispatches lacking shared authorization middleware
  • Comparing authorization behavior across message, callback, webhook, and interaction handlers
  • Tests that exercise admitted and non-admitted users through every supported event type
  • Policy enforcement requiring authenticated user identity and allowlist checks before agent execution
  • Runtime monitoring for agent commands originating from interactive callbacks or unexpected users

Automated code review tools may help identify authorization checks that are missing or inconsistently applied.

Sources

SourceLink
github.comgithub.com/chenhg5/cc-connect
github.comgithub.com/chenhg5/cc-connect/blob/v1.5.0/platform/feishu/feishu.go#L661-L680
github.comgithub.com/chenhg5/cc-connect/issues/1852
vulncheck.comvulncheck.com/advisories/cc-connect-through-1.5.0-user-allowlist-bypass-via-feishu-card-actions

Frequently asked

questions

Not necessarily. Exploitation generally requires the ability to trigger an

interactive card action in a Feishu chat that cc-connect admits, plus an

identity that is not otherwise authorized by the per-user allowlist.

No direct host access is necessarily required. The relevant access is the

ability to interact with the Feishu integration and reach the vulnerable

callback path.

Apply the same centralized per-user authorization decision to every event

path, including interactive card callbacks, immediately before dispatching

agent commands.

Maintain one authorization policy for all interaction types, test each handler

with authorized and unauthorized identities, use deny-by-default behavior,

and limit the privileges available to dispatched agents.

Find out what's already
exploitable in your codebase.