II. Threat categories · Ε
Ε. Tool Poisoning
Poisoned tool descriptions and malicious tool responses
- Rule files
- 118
- Effective
- 116, loaded by at least one lane
- Enforce lane
- 13, stable only
- Download
- rules/index.csv
Most frequent crosswalks
The number in brackets is how many rules in this category carry that item. ATR wrote these mappings; the framework owners have not reviewed or endorsed them.
MITRE ATLAS
- AML.T0049 - Exploit Public-Facing Application(31)
- AML.T0053 - AI Agent Tool Invocation(30)
- AML.T0051.001 - Indirect(21)
- AML.T0051 - LLM Prompt Injection(16)
- AML.T0053 - LLM Plugin Compromise(13)
- AML.T0040 - AI Model Inference API Access(11)
OWASP Agentic Top 10
- ASI05:2026(62)
- ASI02:2026(42)
- ASI04:2026(36)
- ASI03:2026(17)
- ASI01:2026(15)
- ASI06:2026(12)
Rules
| ID | Title | Severity | Maturity | Lanes |
|---|---|---|---|---|
| ATR-2026-00010 | Malicious Content in MCP Tool Response | critical | test | alert · hunt |
| ATR-2026-00013 | SSRF via Agent Tool Calls | critical | test | alert · hunt |
| ATR-2026-00095 | MCP Tool Supply Chain Poisoning | critical | test | none, inert |
| ATR-2026-00096 | Skill Registry Poisoning and Compromised Tool Distribution | critical | test | none, inert |
| ATR-2026-00103 | Hidden LLM Safety Bypass Instructions in Tool Descriptions | critical | test | alert · hunt |
| ATR-2026-00161 | MCP Tool Description — IMPORTANT Tag Cross-Tool Shadowing Attack | critical | test | alert · hunt |
| ATR-2026-00210 | Flowise System Message Override via Template Interpolation (CVE-2025-59528) | critical | test | alert · hunt |
| ATR-2026-00277 | ECHO Template / Jinja / SQL Command Injection via LLM | critical | test | alert · hunt |
| ATR-2026-00415 | Flowise Custom MCP STDIO Command Injection (CVE-2026-40933) | critical | test | alert · hunt |
| ATR-2026-00419 | Cursor MCP JSON Zero-Click Configuration RCE (CVE-2025-54136) | critical | test | alert · hunt |
| ATR-2026-00434 | mcp-remote authorization_endpoint OS Command Injection (CVE-2025-6514) | critical | stable | enforce · alert · hunt |
| ATR-2026-00494 | SQL Injection and Code Injection Attack Payload Detection | critical | test | alert · hunt |
| ATR-2026-00521 | Shell Command Injection in Agent Tool Context | critical | test | alert · hunt |
| ATR-2026-00526 | Claude Code Shell Metacharacter in Double-Quoted File Path | critical | test | alert · hunt |
| ATR-2026-00529 | LiteLLM Proxy SQL Injection (CVE-2026-42208, CISA KEV 2026-05-08) | critical | test | alert · hunt |
| ATR-2026-00530 | ModelScope MS-Agent Shell Tool Unsanitized Argv RCE (CVE-2026-2256) | critical | test | alert · hunt |
| ATR-2026-00531 | PraisonAI Unauthenticated Agent API Exploitation (CVE-2026-44338) | critical | test | alert · hunt |
| ATR-2026-00532 | Apache Doris MCP Server SQL Injection (CVE-2025-66335) | critical | test | alert · hunt |
| ATR-2026-00533 | Apache Pinot MCP Unauthenticated Remote Cluster Takeover | critical | test | alert · hunt |
| ATR-2026-00536 | nginx-ui MCP Endpoint Unauthenticated Command Execution (CVE-2026-33032) | critical | test | alert · hunt |
| ATR-2026-00538 | LangChain-ChatChat Unauthenticated MCP STDIO Server Configuration RCE (CVE-2026-30617) | critical | draft | hunt |
| ATR-2026-00540 | PraisonAI parse_mcp_command() CLI Argument Command Injection (CVE-2026-34935) | critical | draft | hunt |
| ATR-2026-00544 | PraisonAI MCP Path-Traversal .pth Injection RCE (GHSA-9mqq-jqxf-grvw) | critical | draft | hunt |
| ATR-2026-00545 | PraisonAI tool_override.py Unauthenticated RCE — CVE-2026-40287 Patch Bypass (CVE-2026-44334) | critical | draft | hunt |
| ATR-2026-00561 | FastMCP vulnerable to windows command injection in FastMCP Cursor installer via server_name | critical | test | alert · hunt |
| ATR-2026-00572 | SymJack — Symlink Approval-Path Spoofing Redirects Writes into Agent MCP/Config (RCE on Restart) | critical | test | alert · hunt |
| ATR-2026-00575 | Miasma / Phantom Gyp — npm Worm Backdoors AI-Agent Config Files (binding.gyp install-exec + auto-run config injection) | critical | test | alert · hunt |
| ATR-2026-00576 | Hades / Shai-Hulud — AI-Agent Credential Harvester in Supply-Chain Package (Anthropic / Claude / MCP key theft + exfil) | critical | experimental | hunt |
| ATR-2026-00577 | Command Injection in create-mcp-server-stdio via Unsafe exec() Concatenation (CVE-2025-54994) | critical | experimental | hunt |
| ATR-2026-00715 | Tool Knowledge Hijack — Identity Injection with Tool Call Arguments | critical | stable | enforce · alert · hunt |
| ATR-2026-01300 | MCP Tool Description — Notes Parameter Chat-History Exfiltration | critical | test | alert · hunt |
| ATR-2026-01302 | Fake Tool Result Prefix — Injected Instruction via Simulated Completion | critical | stable | enforce · alert · hunt |
| ATR-2026-01306 | MCP OAuth Authorization URL — Command Injection via URL Authority | critical | stable | enforce · alert · hunt |
| ATR-2026-01307 | MCP DNS Rebinding Attack — Hostname Time-Based IP Switching | critical | stable | enforce · alert · hunt |
| ATR-2026-01310 | MCP Tool Description — Compliance/Audit Framing for Mandatory Chat Context | critical | stable | enforce · alert · hunt |
| ATR-2026-01931 | gemini-mcp-tool execAsync Command Injection & @file Exfiltration (CVE-2026-0755) | critical | test | alert · hunt |
| ATR-2026-01935 | LiteLLM Custom-Code Guardrail Sandbox Escape (CVE-2026-40217) | critical | test | alert · hunt |
| ATR-2026-01952 | PraisonAI codeMode JS Sandbox Escape RCE via new Function/with() (GHSA-p69m-4f92-2v84) | critical | test | alert · hunt |
| ATR-2026-01953 | npm PraisonAI codeMode Sandbox Escape via Function Constructor Prototype Chain (GHSA-vmmj-pfw7-fjwp) | critical | test | alert · hunt |
| ATR-2026-01959 | OpenHuman Shell Tool Allowlist Bypass via Env-Prefix / find -execdir (CVE-2026-55743) | critical | test | alert · hunt |
| ATR-2026-01963 | PraisonAI Action Orchestrator step.target Path Traversal Arbitrary File Write RCE (CVE-2026-39305 / GHSA-jfxc-v5g9-38xr) | critical | test | alert · hunt |
| ATR-2026-01965 | Flowise Custom MCP node-load-method OS Command RCE (CVE-2025-8943) | critical | test | alert · hunt |
| ATR-2026-01967 | DeepChat Mermaid XSS to RCE via Electron IPC MCP Server Registration (CVE-2025-66481 / GHSA-h9f5-7hhf-fqm4) | critical | test | alert · hunt |
| ATR-2026-01968 | DeepChat Markdown Deeplink shell.openExternal Protocol Bypass RCE (CVE-2026-43899, GHSA-cp8j-jx7q-7r5f) | critical | test | alert · hunt |
| ATR-2026-01970 | PraisonAI FileTools _validate_path normpath Path Traversal (CVE-2026-35615 / GHSA-693f-pf34-72c5) | critical | test | alert · hunt |
| ATR-2026-01973 | AnythingLLM Logo Endpoint Path Traversal File Read/Delete (CVE-2024-3025) | critical | test | alert · hunt |
| ATR-2026-01978 | AnythingLLM collector /process filename Path Traversal Arbitrary File Deletion (CVE-2023-5832) | critical | test | alert · hunt |
| ATR-2026-01979 | PandasAI Interactive Prompt Injection -> Python Sandbox Escape RCE (CVE-2024-12366 / GHSA-vv2h-2w3q-3fx7) | critical | test | alert · hunt |
| ATR-2026-01980 | Agentic-Flow MCP Tool-Parameter OS Command Injection (GHSA-vcv2-r9jh-99m5) | critical | test | alert · hunt |
| ATR-2026-01983 | MCP-for-Stata: Command Injection via log_file_name Parameter (CVE-2026-47708) | critical | test | alert · hunt |
| ATR-2026-01985 | MCP Connect: Unauthenticated /bridge Endpoint Arbitrary Process Spawn RCE (GHSA-wvr4-3wq4-gpc5) | critical | test | alert · hunt |
| ATR-2026-01987 | Langroid SQLChatAgent Prompt-to-SQL Remote Code Execution (CVE-2026-25879) | critical | test | alert · hunt |
| ATR-2026-02021 | MCP Inspector Unauthenticated Proxy stdio Command Execution (CVE-2025-49596) | critical | test | alert · hunt |
| ATR-2026-02027 | Diagnostic Content Remediation Command Injection (Agentjacking) | critical | test | alert · hunt |
| ATR-2026-02041 | Shell Command Injection via Echo-Marker Proof-of-Concept in Tool Argument | critical | experimental | hunt |
| ATR-2026-02141 | Unsandboxed Command Execution via Dynamic MCP Server Config (command/args Injection) | critical | experimental | hunt |
| ATR-2026-02145 | Python Sandbox Escape via Generator/Coroutine Frame Object Introspection | critical | experimental | hunt |
| ATR-2026-02194 | Malicious Go init() Function Spawning a Process via a Code-Generation Tool | critical | experimental | hunt |
| ATR-2026-00011 | Instruction Injection via Tool Output | high | test | alert · hunt |
| ATR-2026-00012 | Unauthorized Tool Call Detection | high | experimental | hunt |
| ATR-2026-00100 | Consent Bypass via Hidden LLM Instructions in Tool Descriptions | high | test | alert · hunt |
| ATR-2026-00101 | Trust Escalation via Authority Override Instructions | high | test | alert · hunt |
| ATR-2026-00105 | Silent Action Concealment Instructions in Tool Descriptions | high | test | alert · hunt |
| ATR-2026-00106 | Schema-Description Contradiction Attack | high | test | alert · hunt |
| ATR-2026-00209 | MCPwn Runaway Tool Invocation via Retry Directive (CVE-2026-33032) | high | test | alert · hunt |
| ATR-2026-00259 | ANSI Escape Code Terminal Injection | high | test | alert · hunt |
| ATR-2026-00270 | XSS Payload Injection in Tool Response Output | high | test | alert · hunt |
| ATR-2026-00393 | ANSI Code Elicitation Request | high | test | alert · hunt |
| ATR-2026-00435 | Azure MCP Server Missing Authentication for Critical Function (CVE-2026-32211) | high | stable | enforce · alert · hunt |
| ATR-2026-00448 | Spring AI MilvusVectorStore Filter Expression Injection (CVE-2026-41705) | high | test | alert · hunt |
| ATR-2026-00513 | Package Hallucination Exploitation — AI-Suggested Fake Package Installation | high | stable | enforce · alert · hunt |
| ATR-2026-00522 | SQL Injection via Natural Language Agent Interface | high | test | alert · hunt |
| ATR-2026-00534 | Alibaba RDS MCP Unauthenticated Database Metadata Exfiltration | high | test | alert · hunt |
| ATR-2026-00537 | FastMCP Windows cmd.exe Injection via Server Name Metacharacters (CVE-2025-64340) | high | draft | hunt |
| ATR-2026-00541 | Agent Zero MCP Configuration Command Injection via mcp_servers field (CVE-2026-30624) | high | draft | hunt |
| ATR-2026-00542 | Upsonic MCP Command Allowlist Bypass RCE (CVE-2026-30625) | high | draft | hunt |
| ATR-2026-00543 | LiteLLM MCP Server Creation Authenticated argv Injection (CVE-2026-30623) | high | draft | hunt |
| ATR-2026-00567 | MCP stdio server config command injection via unvalidated test endpoints | high | test | alert · hunt |
| ATR-2026-00568 | Agent SSRF to cloud metadata / file inclusion via unvalidated fetch URL | high | test | alert · hunt |
| ATR-2026-00581 | MCP Tool Rug-Pull — Post-Approval Description Redefinition Injects Execution Instructions | high | test | alert · hunt |
| ATR-2026-00714 | Tool Camouflage — Forced Specific Tool Invocation via Injected Instruction | high | stable | enforce · alert · hunt |
| ATR-2026-00716 | Indirect PI — Data Destruction Instruction via Agent File/Document Tool | high | stable | enforce · alert · hunt |
| ATR-2026-01301 | MCP Tool Description — Exclusive Tool Invocation Override | high | stable | enforce · alert · hunt |
| ATR-2026-01303 | Tool Schema Enumeration via Social Engineering | high | stable | enforce · alert · hunt |
| ATR-2026-01775 | MCP Tool-Manifest Poisoning — Name Squatting, Result Shadowing & Covert-Action Directives (Semantic) | high | test | alert · hunt |
| ATR-2026-01927 | mcp-server-kubernetes Command Injection in kubectl_scale / kubectl_patch / explain_resource (CVE-2025-53355) | high | test | alert · hunt |
| ATR-2026-01928 | Framelink Figma MCP Server curl-Fallback Command Injection (CVE-2025-53967) | high | experimental | hunt |
| ATR-2026-01930 | MCP Sampling Prompt Injection (Server-to-Client createMessage Abuse) | high | test | alert · hunt |
| ATR-2026-01932 | Shadow / Undeclared MCP Server Registration (MCP-38: MCP-18) | high | test | alert · hunt |
| ATR-2026-02022 | CurXecute — Cursor .cursor/mcp.json Injected-Server Auto-Exec RCE (CVE-2025-54135) | high | test | alert · hunt |
| ATR-2026-02023 | EscapeRoute — Filesystem MCP Server Directory Prefix-Bypass (CVE-2025-53110) | high | test | alert · hunt |
| ATR-2026-02024 | EscapeRoute — Filesystem MCP Symlink Escape to LaunchAgent Persistence (CVE-2025-53109) | high | test | alert · hunt |
| ATR-2026-02025 | MCP Full Schema Poisoning — Injected Directive in Non-Description inputSchema Field (MCP-11) | high | test | alert · hunt |
| ATR-2026-02102 | HTML/Script Injection in Tool Call Argument Targeting a Human-Approval Dashboard | high | experimental | hunt |
| ATR-2026-02193 | SSRF via Non-Canonical IPv6 Encoding of Loopback/Internal Addresses | high | experimental | hunt |
| ATR-2026-02233 | Unscoped Destructive or Mass-Disclosure Database Operation Request via Natural Language | high | experimental | hunt |
| ATR-2026-02260 | Remediation-Framed Command Execution in Tool Response (Agentjacking) | high | experimental | hunt |
| ATR-2026-02263 | SQL Injection via Unparameterized Template-Expression Value in Workflow-Automation SQL Node (CVE-2026-59257) | high | test | alert · hunt |
| ATR-2026-02350 | MCP JSON-RPC Message Carries Case-Duplicate name/arguments Keys to Smuggle an Unauthorized Tool Call | high | experimental | hunt |
| ATR-2026-02352 | Email Search/Reply Tool Argument Breaks Out of IMAP SEARCH Quoted String to Inject IMAP Commands | high | experimental | hunt |
| ATR-2026-02372 | Unsigned Automated Task/Validation Feed Invokes npm install / npx -y (Lifecycle-Script RCE) | high | experimental | hunt |
| ATR-2026-02376 | MCP Tool Description Defines a Common-Phrase Trigger to Forward Full Conversation History | high | experimental | hunt |
| ATR-2026-02403 | MCP Tool Returns Untrusted External Content Carrying Hidden Agent Instructions Without Spotlighting | high | test | alert · hunt |
| ATR-2026-02557 | Shell Command Separator Inside a Path-Typed MCP Tool Argument | high | test | alert · hunt |
| ATR-2026-02666 | Trusted Hostname Parked in the URL Userinfo So the Real Host Is Whatever Follows the At-Sign | high | test | alert · hunt |
| ATR-2026-02668 | Branch or Tag Ref Name Carrying Shell Command Substitution in Tool Output | high | test | alert · hunt |
| ATR-2026-02707 | Mail or Attachment Tool Argument Splits into a New Header via Embedded CRLF | high | test | alert · hunt |
| ATR-2026-02817 | Untrusted Tool Result Dictates a Tool Call by Schema Parameter Name | high | test | alert · hunt |
| ATR-2026-02864 | Non-Canonical Numeric IPv4 Host in a URL-Typed Tool Argument (SSRF Filter Bypass) | high | test | alert · hunt |
| ATR-2026-02868 | SSRF via IPv6 Transition Address Embedding a Private IPv4 Target | high | test | alert · hunt |
| ATR-2026-02870 | Invisible Unicode Control Before a Directive in a Tool Description or Skill File | high | test | alert · hunt |
| ATR-2026-00720 | Tool Misuse — Privilege Escalation via Social Engineering Agent to Grant Access | medium | stable | enforce · alert · hunt |
| ATR-2026-01982 | dbt-mcp node_selection/resource_type Argument Injection (CVE-2026-44968) | medium | test | alert · hunt |
| ATR-2026-02514 | MCP Server Descriptor URL Field Carries an HTML/Script Breakout Payload | medium | test | alert · hunt |
| ATR-2026-02542 | Observability Query Tool Argument Appends a Pipeline Stage and Comments Out the Rest of the Query | medium | test | alert · hunt |
| ATR-2026-02669 | Query-Language Pipeline Stage Smuggled Into an Identifier-Typed Tool Parameter | medium | test | alert · hunt |
| ATR-2026-02682 | XML/SVG Entity Expansion Bomb in Content Returned to an Agent | medium | test | alert · hunt |
| ATR-2026-02865 | Read-Only Allow-Listed Utility Invoked With a File-Write or State-Changing Option in a Command Argument | medium | test | alert · hunt |
Page generated 2026-11-10Something wrong or missing? Open an issue
