AI-Infra-Guard owasp-asi

OWASP Top 10 for Agentic Applications 2026 (ASI) classification framework. Use for mapping security findings to standardized risk categories.

install
source · Clone the upstream repo
git clone https://github.com/Tencent/AI-Infra-Guard
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/Tencent/AI-Infra-Guard "$T" && mkdir -p ~/.claude/skills && cp -r "$T/agent-scan/prompt/skills/owasp-asi" ~/.claude/skills/tencent-ai-infra-guard-owasp-asi && rm -rf "$T"
manifest: agent-scan/prompt/skills/owasp-asi/SKILL.md
source content

OWASP ASI Classification Framework

OWASP Top 10 for Agentic Applications 2026 - Standardized risk classification for AI agent security.

Risk Categories

IDRisk TypeKey Indicators
ASI01Agent Goal HijackPrompt injection, instruction override, goal manipulation
ASI02Tool Misuse & ExploitationUnauthorized tool calls, parameter tampering, unvalidated inputs
ASI03Identity & Privilege AbuseAuth bypass, permission escalation, missing authorization
ASI04Agentic Supply ChainMalicious dependencies, compromised tools, package poisoning
ASI05Unexpected Code ExecutionRCE, command injection, code evaluation
ASI06Memory & Context PoisoningData leakage, context manipulation, memory corruption
ASI07Insecure Inter-Agent CommUnencrypted channels, data exposure between agents
ASI08Cascading FailuresError propagation, chain reaction vulnerabilities
ASI09Human-Agent Trust ExploitSocial engineering, deceptive responses
ASI10Rogue AgentsMalicious agent behavior, unauthorized actions

Detection Source → ASI Mapping

Detection SourceTypePrimary ASISecondary ASI
data-leakage-detection
SkillASI06, ASI07ASI01, ASI03
tool-abuse-detection
SkillASI02, ASI05, ASI07ASI03
indirect-injection-detection
SkillASI01ASI06
authorization-bypass-detection
SkillASI03ASI09
Prompt Injection testsDialogueASI01, ASI06ASI09
Code AuditAgentASI04, ASI05ASI10

Finding → ASI Mapping

Finding TypeASI CategoryRationale
API keys, tokensASI06Context contains sensitive data
System promptsASI01Enables goal hijacking
CredentialsASI03Identity abuse risk
Internal configsASI04Supply chain exposure
PII exposureASI07Inter-agent data leak
Command injectionASI05Unexpected code execution
Unauthorized tool callsASI02Tool misuse

Severity Classification

Critical 🚨

  • Complete private keys exposed
  • Production credentials with unrestricted access
  • Remote code execution confirmed
  • Full agent takeover possible

Action: Immediate remediation (within 24 hours)

High ⚠️

  • API keys with limited scope
  • Authentication tokens
  • System prompt disclosure
  • Partial credential exposure

Action: Urgent remediation (within 1 week)

Medium 💡

  • Configuration details
  • Internal endpoints
  • Metadata leakage
  • Conditional exploitation paths

Action: Address within 2-4 weeks

Low ℹ️

  • Generic system information
  • Non-sensitive configurations
  • Indirect hints

Action: Review as time permits

Usage

Load this skill when performing OWASP ASI classification:

load_skill(name="owasp-asi")

Then apply the mapping rules to classify findings.