Airlock: Bounding the Multi-Agent Blast Radius via Selective Online Containment
Organizations: Imperial College London · University of Oxford · Independent Researcher
Abstract
When multi-agent systems act on the external world, a single agent's faulty output can spread across tasks before detection, creating a system-wide blast radius. Single-agent guardrails are confined to individual harnesses, while existing multi-agent controls either cancel only the faulty source and miss downstream consumers or halt and roll back the entire workflow at the cost of healthy work. We present Airlock, a multi-agent runtime and guardrail for selective post-alert fault containment. Its tree-structured runtime, EdgeCitadel, narrows permissions vertically from parent to child agents to establish node boundaries and control switches. Meanwhile, an append-only provenance DAG records cross-agent data consumption horizontally. Upon an alert, Airlock traces forward on this DAG to isolate tainted executions and pending actions, and backward to narrow root causes. At the site switches, a default full quarantine tier freezes affected branches without a solver, while an assured optimized cut tier uses fence-then-prune to enforce a deadline-aware minimum cut minimizing collateral disruption. Across 1,832 testbed runs, 161 physical smart-home LLM runs, and 24,266 simulation records, Airlock blocks all post-alert faulty actuations on our main testbed and smart-home grids with zero post-convergence leaks. Whereas native controls in A2A and LangGraph either contain at most 33.3% of faults or retain at most 16% of work without redo, Airlock retains +68 pp to +96 pp more healthy work, and its optimized cut retains +46 pp more than full quarantine under dense spread. These results show that co-designing runtime authority boundaries with online provenance enables surgical fault containment without sacrificing healthy work (code: https://github.com/EdgeCitadelTeam/EdgeCitadel).
Figures & tables
Appendix figures & tables5 assets
Supplementary material from the paper’s appendix.
Appendix
| Action | Site | Port | |
|---|---|---|---|
| task hold on | host( ) | all pending out-edges of | |
| egress hold on | gateway | immediate effect edges out of | |
| revoke | verifier | effect edges out of under | |
| filter | ingress( ) | effect and scheduled edges under | |
| admission block at | ingress( ) | all effect and scheduled edges into | |
| scheduled-effect, result or admission hold on | scheduler, host or admitting site | either |
| S1 | Liveness. A partitioned or lagging site can prevent from arriving; fail-closed site leases keep the site safe by holding new and pending effects upon expiry, at the cost of liveness. |
|---|---|
| S2 | Executed effects. Committed external effects are not rolled back; Theorem 5 governs commits after . |
| S3 | Pre-containment effects. Effects that commit between the alert and (or are accepted under the gateway lock before ) are counted as pre- commits, separate from post- safety. |
| S4 | Declared sources. Safety is relative to the declared initial source set . |
| S5 | Cost optimality. is cost-optimal for the snapshot and mask model; any valid cut is equally safe under Theorem 5. |
| S6 | Mediation. External effects must pass through a site or gateway that enforces . |
| S7 | Masks vs. fences. Deadline masks are planning predictions; convergence and release rely on fenced journal positions. |
| ID | Prediction | Registered | Outcome | Main text |
|---|---|---|---|---|
| H1 | With rogue spread, the optimized cut contains more runs than Trace & stop | by Oct 22 | Not supported: both clean in every rogue run of the main grid (12 and 12 of 12) | Table 3 |
| H2 | Without rogue spread, Trace & stop keeps at least as much work | by Oct 22 | Not supported: Trace & stop keeps less | Table 3 |
| H3 | As control latency grows, Cut without deadlines leaks more than the optimized cut | by Oct 22 | Holds weakly: one run leaked at 4 s (11 of 12 clean), none of the optimized cut’s | §5.5 |
| H4 | The optimized cut issues fewer control operations than Greedy cut | by Oct 22 | Not supported: 16 vs. 17 logical actions, 26.5 vs. 25.5 messages (medians) | Table 3 |
| H5 | stays flat with live workload while grows | by Oct 22 | Holds (sim.): over 40 workflow instances grows from 310 to 8,164 nodes, while has 371 and then 311 | — |
| H6 | Sink guard contains about as much but keeps less work | Oct 21 | Holds (sim.) | — |
| Framework | Stop or cancel | Fence before an effect | Effects already under way | Rollback or replay |
|---|---|---|---|---|
| A2A 1.0.x, a2a-sdk 1.2.3 ( A2A Project, 2026 ) | Cancel one task; no cancel-all (the operator lists and cancels tasks per server) | None for the operator; an agent may reject a task itself | Cancellation is attempted, “but success is not guaranteed” | None |
| LangGraph 1.2.16 ( LangChain, 2026 ) | Cancel a run (interrupt), also in bulk | interrupt() or interrupt-before/after per node; edit the state, then resume | Stops the run’s worker; a tool call may have started | Rollback of graph state; time travel re-runs every later node |
| AutoGen 0.7.5 ( Microsoft, 2026a ) | Cancellation token per run; termination conditions after the current turn; stop the runtime | Approval of code execution; an intervention handler can drop a message before delivery | Async tools are cancelled and code executors killed; sync tools and the current message finish | Save and load of state |
| CrewAI 1.15.28 ( CrewAI, 2026 ) | No stop or cancel; time and iteration limits | Before-tool-call hook (other hook errors fail open) | Guardrails and human review run after execution | Checkpoints; replay re-runs every later task |
| OpenAI Agents SDK 0.24.0 ( OpenAI, 2026b ) | Cancel a streamed run, now or after the turn; turn limit | Tool approval; tool input guardrail (function tools only); blocking input guardrail | Async tools are cancelled at their next await; sync tools are not stopped | Restore of run state or session history |
| MCP 2026-07-28, SDK 2.3.0 ( Model Context Protocol, 2026 ) | Cancel a request or a task | The host confirms tool calls; consent is advisory | The receiver may ignore a cancel and “is not obligated to actually stop the work” | None (stateless) |
AI Agent Safety
AI agent safety studies how to prevent and detect harmful actions by autonomous systems that use tools, memory, and multi-step plans.
See all 349 papers