OpenClaw's Conflicting Tool Identity Hints Bypass Dangerous-Tool Prompting

Description

Summary

ACP permission resolution trusted conflicting tool identity hints from rawInput and metadata, which could suppress dangerous-tool prompting.

Affected Packages / Versions

  • Package: openclaw (npm)
  • Affected: < 2026.3.22
  • Fixed: >= 2026.3.22
  • Latest released tag checked: v2026.3.23-2 (630f1479c44f78484dfa21bb407cbe6f171dac87)
  • Latest published npm version checked: 2026.3.23-2

Fix Commit(s)

  • e4c61723cd2d530680cc61789311d464ab8cdf60

Release Status

The fix shipped in v2026.3.22 and remains present in v2026.3.23 and v2026.3.23-2.

Code-Level Confirmation

  • src/acp/client.ts now fails closed when meta, rawInput, and title tool identities conflict instead of trusting spoofable raw input.
  • src/acp/client.test.ts ships regressions for conflicting tool identity hints and dangerous-tool prompting.

OpenClaw thanks @zpbrent for reporting.

Basic information

Type
reviewed
Severity
medium
Advisory on GitHub
Open advisory ↗
Repository advisory
Open repository advisory ↗
Source code
Browse source ↗
Published (advisory)
2026-03-26 21:46:42 UTC
Updated
2026-04-10 19:42:12 UTC
GitHub reviewed
2026-03-26 21:46:42 UTC
NVD published
2026-04-10

EPSS Score

Score Percentile
0.04% 12.08%

CVSS Scores

Base score Version Severity Vector
5.7 3.1
CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:U/C:N/I:H/A:N Click to expand
Attack vector (AV:N)
Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L)
Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:L)
A normal user session is enough; they don’t have to be admin.
User interaction (UI:R)
A real person has to do something—click, install, enable—otherwise it doesn’t land.
Scope (S:U)
Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems.
Confidentiality (C:N)
Doesn’t really leak secrets in a meaningful way.
Integrity (I:H)
They could widely tamper with or forge data—trust in the data is badly hurt.
Availability (A:N)
Service keeps running; no real outage angle.
6.9 4.0
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:P/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N Click to expand
Attack vector (AV:N)
Could be attacked over the internet or any normal routed network.
Attack complexity (AC:L)
Exploitation conditions are straightforward and stable.
Attack requirements (AT:N)
No additional preconditions are required beyond normal reachability.
Privileges required (PR:L)
Low privileges are required.
User interaction (UI:P)
A user has to participate (for example click/open/approve).
Vulnerable system confidentiality impact (VC:N)
No confidentiality impact on the vulnerable system.
Vulnerable system integrity impact (VI:H)
High integrity impact on the vulnerable system.
Vulnerable system availability impact (VA:N)
No availability impact on the vulnerable system.
Subsequent system confidentiality impact (SC:N)
No confidentiality impact on subsequent systems.
Subsequent system integrity impact (SI:N)
No integrity impact on subsequent systems.
Subsequent system availability impact (SA:N)
No availability impact on subsequent systems.

Identifiers

CWEs

CWE id Name
CWE-807 Reliance on Untrusted Inputs in a Security Decision
CWE-863 Incorrect Authorization

Credits

  • zpbrent (reporter)

Affected packages (1)

Vulnerable version ranges and first patched releases as published by GitHub.

Ecosystem Package Vulnerable range First patched Vulnerable functions
npm openclaw < 2026.3.22 2026.3.22

References

cvelogic Threat Intelligence