PulseAugur
EN
LIVE 06:59:06

AI Tool Schema Poisoning Vulnerability Bypasses Content Filters

By PulseAugur Editorial · [1 sources] ·

A new vulnerability, CVE-2025-54136, dubbed "MCP Tool Schema Poisoning," allows attackers to silently alter the functionality of AI tools by manipulating their JSON schema definitions. This attack bypasses traditional content filters because the schema mutation occurs at a structural level, not within the prompt text. Once a tool's schema is approved, subsequent changes can grant malicious capabilities like unauthorized file access or network calls, as demonstrated by the MCPoison vulnerability which achieved a 36.5% success rate across various LLMs. AI

IMPACT Highlights a critical security gap in AI agent frameworks, necessitating new validation mechanisms beyond prompt-level filtering.

RANK_REASON Disclosure of a novel vulnerability and associated benchmark testing in AI tool interaction. [lever_c_demoted from research: ic=1 ai=1.0]

Read on dev.to — MCP tag →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

AI Tool Schema Poisoning Vulnerability Bypasses Content Filters

How we ranked this

Signal score
24 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Disclosure of a novel vulnerability and associated benchmark testing in AI tool interaction. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
safety, product, paper
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. dev.to — MCP tag TIER_1 English(EN) · Davi ·

    MCP Tool Schema Poisoning: When the Framework Layer Authorizes the Attack

    <p>A researcher commits a legitimate MCP server definition to a shared repository. Teammates approve the tool manifest. Two weeks later, the tool schema silently changes on the remote server. Nobody checks again. The tool now accepts a <code>destination</code> parameter that rout…

RELATED ENTITIES

RELATED TOPICS