Skip to Content
Scan your AI agents for free·npx -y @inkog-io/cli scan .·Get Started →
VulnerabilitiesOverview

Vulnerability Database

Browse Inkog’s comprehensive database of AI agent security vulnerabilities.

Each vulnerability includes detection rules, affected frameworks, and remediation guidance.

Scan your agent code
$npx -y @inkog-io/cli scan .
Free tier available · 60s scan · Get Started →

Vulnerability Categories

Injection Attacks

Vulnerabilities where untrusted input manipulates AI behavior:

  • Prompt Injection: User input alters LLM instructions
  • Indirect Prompt Injection: Malicious content in retrieved data
  • Template Injection: Dynamic template construction with user data

Authorization & Access Control

Improper access controls in agent systems:

  • Tool Use Without Validation: Agents execute tools without input checks
  • Privilege Escalation: Agents perform actions beyond intended scope
  • Missing Authentication: Unprotected agent endpoints

Memory & State

Vulnerabilities in agent memory and state management:

  • Memory Poisoning: Malicious data persisted to long-term memory
  • State Manipulation: Session state altered by untrusted input
  • Context Window Attacks: Exceeding context limits to truncate instructions

Information Disclosure

Unintended exposure of sensitive information:

  • Chain-of-Thought Leakage: Internal reasoning exposed to users
  • System Prompt Extraction: Attackers extract hidden instructions
  • Sensitive Data in Prompts: Credentials or PII in prompt text

Code Execution

Risks from dynamic code execution:

  • Unrestricted Code Interpreters: Sandbox escapes in code tools
  • Unsafe Deserialization: Arbitrary object instantiation
  • Command Injection: Shell commands with user input

Severity Levels

LevelDescriptionResponse
CriticalImmediate exploitation possible, severe impactFix immediately, block deployment
HighLikely exploitable with significant impactFix before next release
MediumExploitable under specific conditionsFix in normal development cycle
LowLimited exploitability or impactTrack and address as time permits

OWASP LLM Top 10 Mapping

Inkog rules map to the OWASP Top 10 for LLM Applications . Not every category is detected; see the status definitions on the coverage page.

OWASP GenAI LLM Top 10 2026. Full status definitions and notes: OWASP LLM Top 10 coverage.

OWASP LLMInkog rules
LLM01: Prompt Injectionprompt_injection, prompt_template, skill_tool_poisoning
LLM02: Sensitive Information Disclosurecross_tenant, hardcoded_credentials, logging_sensitive_data, pii_filter_wiring, skill_cross_origin, unsafe_env_access
LLM03: Excessive Agencyexcessive_permissions, goal_conflict_detection, missing_authz, missing_oversight, skill_autonomy_abuse, skill_excessive_permissions, token_validation
LLM04: Supply Chainskill_supply_chain, supply_chain, unsafe_deserialization
LLM05: Data and Model Poisoningdata_poisoning
LLM06: Unbounded Consumptioncontext_exhaustion, infinite_loop, missing_rate_limits, recursive_tool_calling, token_bombing
LLM07: Misinformationoverreliance
LLM08: Hidden Context Exposuresystem_prompt_leak
LLM09: Vector and Embedding Weaknessesrag_overfetching
LLM10: Improper Output Handlingcommand_injection, exec_eval, output_validation, path_traversal, sql_injection, ssrf

OWASP Top 10 for Agentic Applications

Rules also carry OWASP Top 10 for Agentic Applications 2026  ids, delivered in compliance_mapping.owasp_agentic_threats:

OWASP Top 10 for Agentic Applications 2026. Generated from the rule engine’s metadata; do not edit by hand.

CategoryStatusRules that stamp this id
ASI01: Agent Goal HijackActive detectiongoal_conflict_detection, prompt_injection, prompt_template, skill_tool_poisoning
ASI02: Tool Misuse and ExploitationActive detectionoutput_validation, path_traversal, sql_injection, ssrf
ASI03: Identity and Privilege AbuseActive detectioncross_tenant, excessive_permissions, hardcoded_credentials, missing_authz, skill_excessive_permissions, token_validation, unsafe_env_access
ASI04: Agentic Supply Chain VulnerabilitiesActive detectionskill_supply_chain, supply_chain
ASI05: Unexpected Code ExecutionActive detectioncommand_injection, exec_eval, unsafe_deserialization
ASI06: Memory and Context PoisoningActive detectiondata_poisoning, rag_overfetching
ASI07: Insecure Inter-Agent CommunicationActive detectionskill_cross_origin, unsigned_messages
ASI08: Cascading FailuresActive detectioncontext_exhaustion, infinite_loop, missing_rate_limits, recursive_tool_calling, token_bombing
ASI09: Human-Agent Trust ExploitationActive detectionoverreliance, skill_autonomy_abuse
ASI10: Rogue AgentsActive detectionmissing_oversight

Palo Alto Unit 42 Threat Categories

Inkog detects logic flaws and security risks across the 7 Agentic AI Threat Categories  identified by Palo Alto Unit 42:

CategoryInkog Coverage
1. Prompt InjectionFull coverage
2. Tool MisuseFull coverage
3. Intent BreakingPartial (goal conflict detection)
4. Identity SpoofingFull coverage
5. Unexpected RCEFull coverage
6. Agent Comms PoisoningFull coverage
7. Resource OverloadFull coverage

Rule ids in scan output

Every finding carries a pattern_id (for example exec_eval, sql_injection, missing_oversight). Use it to suppress a finding with an inline comment or a .inkogignore entry; see Configuration. The full list of rule ids, grouped by OWASP category, is on the OWASP coverage page.

Requesting a rule

Detection rules are maintained in Inkog’s engine and ship to every scan without an upgrade. To request a new rule or report a false positive, open an issue on github.com/inkog-io/inkog  with a minimal vulnerable example and, if possible, the safe variant. Custom organization-specific rules are not user-configurable today.

Last updated on