Upgrade to Obot 0.26.2
Immediately upgrade Obot deployments to version 0.26.2 or higher. The patch enforces strict authorization policies across all MCP protocol methods.
Obot AI Agent Platform โ Virtual Model Context Protocol (vMCP) Router & Policy Gateway Traditional CVSS v3.1 scores this vulnerability at 8.8 (High) as an authenticated horizontal authorization bypass across application components. The Hermes Threat Score assesses it at 90 (CRITICAL) with an Agentic Security Score (HASS) of 95 (CRITICAL). In autonomous AI systems, the Model Context Protocol (MCP) serves as the communication standard connecting LLM agents to sensitive knowledge bases, custom tools, and prompt libraries. Obot introduced virtual MCP (vMCP) profiles to enforce tenant isolation and granular access control. However, access policy filters were enforced solely on tool execution (tools/call), omitting validation on prompts, resources, and resource templates. An authenticated user or subagent matching any vMCP profile can extract proprietary system prompts, internal corporate reasoning templates, and confidential vector store resources belonging to other tenants via shared component connections (MITRE ATLAS AML.T0024 & AML.T0040).
Software platform affected by security vulnerabilities and agentic attack patterns.
“Confirmed security vulnerability in Obot Autonomous AI Agent Platform documented in Hermes dossier.”
Obot enables organizations to build, orchestrate, and host autonomous AI agents. To connect agents to external services, Obot relies on the Model Context Protocol (MCP), exposing servers through a virtualized gateway (vMCP) that manages credentials, permissions, and tenant isolation.
| Parameter | Technical Specification | Operational Details |
|---|---|---|
| CVE Identifier | CVE-2026-105139 | International vulnerability catalog identifier |
| Affected Product | Obot AI Agent Platform | Enterprise autonomous multi-agent orchestration runtime |
| Vulnerable Component | pkg/gateway/vmcp/proxy.go & pkg/api/handlers/vmcp.go | vMCP reverse proxy JSON-RPC dispatch controller |
| Vulnerability Class | CWE-285 / CWE-862 / CWE-639 | Missing Authorization / Bypassed Access Control |
| Attack Vector | Network (/api/vmcp/{profile_id}/* JSON-RPC) | Authenticated MCP protocol requests |
| Remediation Release | Obot v0.26.2 | Comprehensive policy validation across all MCP methods |
[ Low-Privilege User / Compromised Agent ] | 1. POST /api/vmcp/{profile_id} JSON-RPC: {"method": "prompts/list"} | v +-------------------------------------------+ | Obot vMCP Gateway Router | | - Checks Tool Call Authorization? | | -> NO (Request is 'prompts/list') | | - Authorization Filter Skipped! | +-------------------------------------------+ | v +-------------------------------------------+ | Shared Component Connection Pool | | - Uses Profile Owner's Full Privileges | | - Queries Upstream MCP Servers | +-------------------------------------------+ | v +-------------------------------------------+ | Confidential Enterprise Assets | | - Tenant System Prompts & Guardrails | | - Proprietary Financial Contexts | | - Restricted Resource Templates | +-------------------------------------------+ | v 2. Attacker exfiltrates cross-tenant data!The vulnerability lies in the JSON-RPC request dispatcher implemented in pkg/gateway/vmcp/proxy.go. When processing incoming requests forwarded from an agent LLM, the handler inspects the JSON-RPC method to decide whether access control rules should be evaluated:
// Vulnerable Implementation in pkg/gateway/vmcp/proxy.gofunc (p *VMCPProxy) HandleRPC(ctx context.Context, profile *types.VMCPProfile, req *jsonrpc.Request) (*jsonrpc.Response, error) { // Only tools/call was subjected to profile policy enforcement if strings.HasPrefix(req.Method, "tools/") { if err := p.enforceToolPolicy(ctx, profile, req); err != nil { return nil, err } }
// Missing check: prompts/* and resources/* were forwarded directly! // The request executes through the component owner's shared client connection conn, err := p.getSharedComponentConnection(profile.ComponentID) if err != nil { return nil, err }
return conn.Forward(ctx, req)}Because strings.HasPrefix(req.Method, "tools/") was the sole gatekeeper, requests using the Model Context Protocol standard methods:
prompts/listprompts/getresources/listresources/readresources/templates/listbypassed the authorization layer completely. The gateway used the shared connection established by the profile owner (often an administrative service identity) to query the upstream MCP server, returning unrestricted results to the unauthorized caller.
In modern agentic systems, MCP resources and prompts represent the intellectual property and boundary controls of the agent:
prompts/get request targeting an internal agent (e.g., executive-finance-advisor). The response discloses the complete system prompt, including proprietary business logic, few-shot demonstration data, and confidential executive guidance.resources/read with a URI pointing to internal documentation or vector index fragments (internal://knowledge/q3-earnings-draft.md). The vMCP proxy retrieves and returns the text payload without validating workspace access.title: Obot vMCP Unauthorized Prompt and Resource Queryid: a8f4b231-1051-39e1-bc84-obotvmcp0001status: stabledescription: Detects JSON-RPC calls querying prompts or resources via Obot vMCP endpoints originating from untrusted client sessions.author: Hermes Codex Cyber Intelligencereferences: - https://github.com/obot-platform/obot/security/advisories/GHSA-2c9v-735x-89q5 - https://vulncheck.com/advisories/obot-0.26.0-before-0.26.2-authorization-bypass-via-vmcp-profile-prompts-and-resourcesdate: 2026-10-10logsource: category: application product: obotdetection: selection_endpoint: http_path|contains: '/api/vmcp/' selection_payload: request_body|contains: - '"method":"prompts/list"' - '"method":"prompts/get"' - '"method":"resources/list"' - '"method":"resources/read"' condition: selection_endpoint and selection_payloadfields: - user_id - http_path - request_bodyfalsepositives: - Legitimate authorized administrative agent debugginglevel: high// Detect anomalous vMCP prompt enumeration in Obot access logsAppServiceHTTPLogs| where CsUriStem has "/api/vmcp/" and CsMethod == "POST"| where RequestBody has "prompts/" or RequestBody has "resources/"| summarize Count = count(), UniqueUsers = dcount(UserPrincipalName), DistinctMethods = make_set(extract('"method":"([^"]+)"', 1, RequestBody)) by ClientIP, CsUriStem, bin(TimeGenerated, 15m)| where Count > 20| order by Count desc#!/usr/bin/env bash# Audit Obot container logs for vMCP prompt/resource requestsdocker logs obot-server 2>&1 | grep -E "(\"method\":\"prompts/|\"method\":\"resources/)" | awk '{print $1, $2, $7, $8}' | sort | uniq -c | sort -nrUpgrade to Obot 0.26.2
Immediately upgrade Obot deployments to version 0.26.2 or higher. The patch enforces strict authorization policies across all MCP protocol methods.
Isolate vMCP Connections
Configure distinct MCP client connection pools per tenant rather than relying on global shared component connections.
Audit Exported Prompts
Review all registered MCP prompt templates to verify that no static credentials, API keys, or unmasked PII are embedded in instructions.
Implement Agent Guardrails
Deploy runtime agent monitors to inspect outbound JSON-RPC responses for unauthorized resource leaks and prompt reflections.