CVE Tools
Track your stack

CVE-2026-65975

Pydantic AI AG-UI Adapter: A dangling client-submitted tool call can execute when a trailing message is dropped during `sanitize_messages`

No known exploitation. EPSS puts it in the 25th percentile. A vendor fix is available.

CVSS 3.16.5MediumEPSS percentile25th0.3% chance in 30 daysCISA KEVNot listedNo confirmed exploitation
No public exploit knownFixVendor fixNo build named by the sources

Published Updated Sources: CVE.org, NVD

NVD CVE.org

What to do

The vendor has published a fix. Version details are below where the sources state them.

  • Pydantic-aiAffected: >= 1.88.0, < 1.107.1Upgrade to: not statedAffected: >= 2.0.0b1, < 2.5.0Upgrade to: not statedStated by: CVE record
  • Pydantic-ai-slimAffected: >= 1.88.0, < 1.107.1Upgrade to: not statedAffected: >= 2.0.0b1, < 2.5.0Upgrade to: not statedStated by: CVE record

Versions as the sources state them. Full version matrix after sign-in

What it is

From the CVE record

Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. In versions 1.88.0 up to but not including 1.107.1 and 2.0.0b1 up to but not including 2.5.0, the UI adapters (AG-UI via Agent.to_ag_ui()/AGUIAdapter, and Vercel AI via VercelAIAdapter) use sanitize_messages to strip unresolved ("dangling") client-submitted tool calls from untrusted message history before it reaches the agent, a defense-in-depth default that prevents the agent from executing tool calls the model never emitted. However, the strip anchored to a message index computed before sanitization ran, so when a trailing client message sanitized to empty and was dropped (for example a client system message under the default manage_system_prompt='server'), a preceding assistant response carrying an unresolved tool call became the new tail and was dispatched without inspection. As a result, a remote client could cause a registered, non-approval server tool to run with client-supplied arguments rather than arguments the model produced. The impact is bounded by what the affected tools do and is most significant for applications that gate tool execution in a model-request hook (before_model_request / after_model_request), since a forged call skips the model turn and bypasses that guardrail; approval-gated tools (requires_approval=True) are not auto-executed by this path. This issue has been fixed in versions 1.107.1 and 2.5.0.

In plain language

No plain-language summary for this CVE yet.

Exploitation

Where each signal puts this CVE on the scale from published to confirmed exploited.

EPSS25th
CISA KEV

Not in the catalog. CISA has not confirmed exploitation.

Public exploits

No public exploit or proof of concept found in the sources we track.

EPSS

0.3% chance of exploitation activity in the next 30 days, which ranks it in the 25th percentile of scored CVEs.

Exploit Prediction Scoring System, FIRST.org. A probability, not a confirmation.

Lifecycle

8 events over 6 days, from the signal feeds we watch.

  1. Patch availablerecord updated
  2. Record updatedrecord updated
  3. Publishedweakness classified, att&ck mapped, record updated

Affected products

Technical detail

CVSS 3.1 vector

Open in the CVSS calculator

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N

Scored 6.5 by NVD.

How it is reached

  • Attack Vector NetworkExploitable remotely over the network without any special conditions
  • Attack Complexity LowNo special conditions — the attack can be reliably reproduced
  • Privileges Required NoneNo authentication required — anyone can exploit this
  • User Interaction NoneNo user interaction needed — fully automated exploitation

Scope

  • Scope UnchangedImpact is limited to the vulnerable component itself

Impact if exploited

  • Confidentiality LowSome restricted information is disclosed, but limited in scope
  • Integrity LowData modification is possible but limited in scope or consequence
  • Availability NoneNo availability impact

Weaknesses

ATT&CK techniques

Mapped from the weaknesses above (CWE to ATT&CK), not observed in attacks.

Sources

References in the record

Watch the software you run.

My Stack ranks new CVEs for your products by real-world exploitation, so the next serious one reaches you without reading every advisory.

We'll flag the next CVE, public exploit or patch for Pydantic Ai, not every advisory.