hunt-agent-multi-agent
Cross-agent trust, privilege, and shared-state abuse in multi-agent systems, including A2A (Agent-to-Agent) protocol exposure. Confused-deputy across agents, privileged-tool reach via a 'clearance'/orchestrator agent, direct A2A/agent-card access, indirect prompt injection through shared context/memory, and shared-state races. Use when a target has >1 agent, an orchestrator/sub-agent split, an A2A endpoint, or tools gated by a 'privileged session'.
elementalsouls · 已有静态检查记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add elementalsouls/Claude-AgentHunter --skill hunt-agent-multi-agent技能说明
name: hunt-agent-multi-agent description: "Cross-agent trust, privilege, and shared-state abuse in multi-agent systems, including A2A (Agent-to-Agent) protocol exposure. Confused-deputy across agents, privileged-tool reach via a 'clearance'/orchestrator agent, direct A2A/agent-card access, indirect prompt injection through shared context/memory, and shared-state races. Use when a target has >1 agent, an orchestrator/sub-agent split, an A2A endpoint, or tools gated by a 'privileged session'." sources: arkx_ctf(field-validated), owasp_agentic_asi_2026, owasp_llm_top10_2025, mitre_atlas report_count: 1 validated_against: "LAB-CONFIRMED (lab/exploits/exploit_multi_agent.py): attacker input forwarded by agent A triggered a privileged escalation in agent B; control: a benign task produced no downstream effect. identity-payload relay CONFIRMED on AgentDojo slack injection_task_5 (agent invited/added/removed an attacker account; URL-indicator probe missed it, taint probe caught it; benchmark under-scored it via case-sensitive name match), 2026-09-12. A2A protocol direct access — vault agent's nonce-response 'auth' completable by any client → TOP SECRET resource (ARKX The Archivist). Confused-deputy / privilege-predicate techniques against a clearance agent (ARKX The Octopus) observed, tagged REFERENCE."
Hunt Agent Multi-Agent Systems
Multi-agent systems create trust boundaries a single agent doesn't have: an orchestrator that trusts sub-agents, a privileged/"clearance" agent others can reach, shared memory, A2A endpoints, and tools that fire "if the session is privileged." Each boundary is a confused-deputy. The winning move is rarely "jailbreak one agent" — it's make a trusted agent take an action you couldn't take directly, or speak the inter-agent protocol yourself.
Where This Lives / Attack Surface Signals
- Agent references "another agent / the files agent / clearance / the orchestrator / escalate to…".
- A tool
read_root_flag/get_system_status/admin_*gated by "if privileged/authorized". - Roles split by agent (utilities / files / clearance).
- Shared scratchpad/memory/cache tool one agent writes and another reads → indirect-injection channel.
- A2A signals: an
.well-known/agent.jsonagent card, an/a2aendpoint,tasks/send, nonce/handshake "auth", "skills"/"clearance" advertised by another agent.
Crown Jewel Targets
Enterprise agent meshes: a low-priv support agent that can reach a high-priv data/ops agent; A2A/agent-card servers exposing privileged "skills"; shared memory poisonable to affect all users; service-account trust that ignores end-user identity (the confused-deputy root).
Characterize First
- Enumerate every agent + every tool (per agent). Map which agent owns the privileged tool and which you can talk to.
- Find the trust edge — who invokes/forwards to whom; does B trust A's assertion ("A says user is authorized")?
- Find the privilege predicate — what "privileged/authorized" actually checks (a shared flag? a peer's claim? a tool result?). That predicate is the target.
- A2A discovery — fetch
.well-known/agent.json, read advertised skills + auth scheme; identify the endpoint (/a2a,tasks/send). - Separate distraction from sink — "multi-agent" is often narrative cover for one exploitable tool.
Technique 1 — A2A protocol direct access [CONFIRMED]
The agent card advertises the protocol; the "auth" is often just protocol steps, not identity.
1. Discover: GET .well-known/agent.json → endpoint, skills, auth scheme (e.g. nonce-response)
2. tasks/send with the request in the correct message format → receive nonce challenge
3. Reply with the same task id + nonce → "ACCESS GRANTED" → read the privileged skill/resource
Field example: a vault agent's nonce-response scheme was completable by any client that followed the protocol → TOP SECRET resource. The nonce is a handshake, not authentication. Root cause: caller identity never verified (no signed token/mTLS).
Technique 2 — Confused deputy across agents [REFERENCE]
Make the privileged agent act because a "trusted" peer asked: relay a request to the clearance agent; assert the privilege predicate through the trusted channel; if an orchestrator forwards structured requests, embed the sub-agent instruction so it's carried verbatim. Verify the gated tool returns the artifact, not just "I asked."
Technique 3 — Privilege-predicate manipulation [REFERENCE]
Attack the predicate, not the gated tool: set/spoof whatever marks the session privileged; mass-assign it ({"privileged":true}/{"role":"admin"}) via a shared-state write; or race set-vs-use (→ hunt-agent-race-condition).
Technique 4 — Indirect injection via shared context/memory [REFERENCE]
Plant the payload where the victim agent reads it: a doc the summarizer ingests; a cache/manifest another agent trusts; poisoned long-term memory (persists for later users — ASI06); ASCII/Unicode-tag smuggling (U+E0000–E007F, invisible to human reviewers). Needs a real sink — a trusted agent taking a real action.
Research anchor: "Prompt Infection: LLM-to-LLM Prompt Injection within Multi-Agent Systems" (Donghyun Lee & Mo Tiwari, arXiv:2410.07283) — a payload that instructs each agent to copy itself into its own output self-replicates across interconnected agents like a worm, spreading even when agents don't share all messages; the paper finds multi-agent systems "highly susceptible" and proposes LLM Tagging as a partial mitigation. This is the difference between one confused-deputy hop (T2) and system-wide propagation — test for the marker re-appearing downstream (see the probe). Maps ASI07/ASI10, ATLAS AML.T0061 LLM Prompt Self-Replication.
Technique 5 — Shared-state race [CONFIRMED pattern]
Multiple agents mutating one balance/quota/flag with no distributed lock → double-spend / limit bypass / privilege flip. See hunt-agent-race-condition.