CVE-2026-121938: langchain4j-aideepin Character IDOR & AI Prompt Exfiltration
A high-severity authorization vulnerability (CVE-2026-121938) was cataloged in langchain4j-aideepin prior to 3.31.0, an enterprise framework powering autonomous agents and personas. The CharacterService.detail function fails to verify whether the requesting user owns the requested character profile. Authenticated attackers enumerating UUIDs via GET /character/{uuid} can extract confidential system prompts, fine-tuned agent instructions, embedded API keys, and multi-tenant conversational histories.
HERMES THREAT SCORE & AGENTIC CONTEXT EXPOSURE EVALUATION
Target:langchain4j-aideepin CharacterService.detail (< 3.31.0) 1. Technical Flaw & Data Leakage Flow
Section titled โ1. Technical Flaw & Data Leakage FlowโIn multi-tenant AI deployments, langchain4j-aideepin allows organizations to define specialized agent personas with unique voice settings, system instructions, and tool integrations. The endpoint GET /character/{uuid} maps directly to:
/* Vulnerable logic prior to version 3.31.0 */public CharacterVO detail(String uuid) { // Missing user ownership check: queries directly by UUID alone! Character character = characterMapper.selectByUuid(uuid); return convertToVO(character);}Because character records store foundational system prompts, safety guardrail overrides, and tool API keys, any authenticated tenant querying foreign UUIDs obtains the full proprietary definition of competing agents, facilitating prompt extraction and jailbreaking attacks (AAP-001 / AAP-002).
2. Remediation Strategy
Section titled โ2. Remediation Strategyโ- Apply Upgrade: Update
langchain4j-aideepinto version3.31.0or higher, which enforces tenant ownership filtersWHERE uuid = ? AND user_id = ?. - Rotate Exposed Secrets: Invalidate all API keys and LLM service tokens embedded within configured character tools.
- Audit Access Logs: Inspect HTTP request logs for sequential or anomalous
GET /character/*enumeration patterns.
CVE-2026-121938: langchain4j-aideepin Character IDOR & AI Prompt ExfiltrationVULNERABILITY
Multi-agent coordination framework and cyclic state graph orchestration engine for tool-calling agents.
๐ Why is this related? (Evidence & Provenance)
“Confirmed security vulnerability in LangGraph Multi-Agent Runtime documented in Hermes dossier.”
- [vulnerability_report]
- [government_confirmation]CISA verified active exploitation in the wild and mandated federal remediation deadline in KEV entry. — Source: Cybersecurity & Infrastructure Security Agency (CISA): CISA Adds CVE-2026-59822 to Known Exploited Vulnerabilities Catalog (Reliability: VERY_HIGH)
Adversary injects explicit formatting tags or role-inversion prompts directly into user input to strip system instructions and escape developer-enforced guardrails.
๐ Why is this related? (Evidence & Provenance)
“CVE-2026-121938 weaponizes the agentic attack pattern formalized under AAP-001.”
- [technical_analysis]Pillar Security demonstrated that executing export BASH_ENV in Auto-Run causes bash to source hostile payloads upon subsequent commands. — Source: Pillar Security Research: Bypassing Cursor Auto-Run: When Shell Built-ins Lead to Host RCE (Reliability: HIGH)
Adversary embeds covert payload instructions into retrieved external data (web pages, repositories, emails) that subvert model planning when parsed by autonomous agents.
๐ Why is this related? (Evidence & Provenance)
“CVE-2026-121938 weaponizes the agentic attack pattern formalized under AAP-002.”
- [technical_analysis]Pillar Security demonstrated that executing export BASH_ENV in Auto-Run causes bash to source hostile payloads upon subsequent commands. — Source: Pillar Security Research: Bypassing Cursor Auto-Run: When Shell Built-ins Lead to Host RCE (Reliability: HIGH)