CVE-2026-133161

Pydantic AI Harness ANSI stripping could block concurrent agent work

pydantic · pydantic-ai-harness

First documented disclosure · Review checkpoint

Summary

Crafted tool output could consume the process's event loop for seconds or minutes.

Severity and exploitation evidence

High · 7.5/10 · CVSS 3.1. Severity source.

Exploitation not established: Malicious exploitation remains unverified in the reviewed originals. A demonstration or absence from KEV does not establish exploitation status.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

The ANSI-removal pattern has quadratic behavior and runs before output size reduction.

Conditions required

ToolOutputLimits strip_ansi=True and attacker-influenced tool output; the default is False.

Potential impact

Other requests and agent runs sharing the process can be delayed.

Affected versions and configurations

pydantic-ai-harness: 0.4.0 through 0.69.0.

Fix and mitigation status

pydantic-ai-harness 0.70.0. Pydantic AI v1 users must migrate to v2 for that harness version or keep strip_ansi disabled.

Practical next steps

Apply the issue-specific correction: pydantic-ai-harness 0.70.0. Pydantic AI v1 users must migrate to v2 for that harness version or keep strip_ansi disabled. Restrict the affected tool or endpoint until the correction is deployed.

Operating lesson

Bound expensive preprocessing before it runs on a shared event loop.

Evidence and limits

Earliest reviewed public evidence 2026-11-18; CVE publication 2026-11-19. CVSS 3.1 7.5 is attributed to Pydantic in the CNA record.

Sources