Softlander Book a call
Agent sandbox assurance

Malicious-code execution testing

Validate controls around potentially harmful generated code

An agent is permitted to execute code obtained from untrusted task content. The decision is how to validate controls around potentially harmful generated code. The work begins with your actual workflow and looks for the evidence that separates intended behaviour from a consequential exception. It is particularly relevant to application security specialists and execution-platform owners. A useful outcome is an explicit account of what was examined, what remains uncertain, and which decision the evidence can support.

Book a 30 minute readiness callBrowse this capability group

The runtime isolation relationship

  1. Starting evidenceExecution policies

    Establish the basis for execution approval.

  2. Decision boundaryRuntime restriction

    Constrain file, process, and network effects.

  3. Decision artefactControlled execution test plan

    Retain the scoped finding and its unresolved dependencies.

Proposed working model. Untested surfaces remain outside the conclusion.

Decision

Frame the decision around execution approval

Questions and the evidence they need

Criterion / evidenceExecution approval
Inspect execution policies for the relevant constraint or observation.
Criterion / evidenceRuntime restriction
Inspect allowed interpreters for the relevant constraint or observation.
Criterion / evidenceObserved behaviour
Inspect network limits for the relevant constraint or observation.
Read each criterion against the available evidence before deciding that additional work is needed.

The decision is how to validate controls around potentially harmful generated code. This is relevant to application security specialists and execution-platform owners when an agent is permitted to execute code obtained from untrusted task content. Name the affected workflow, the consequence of a wrong decision, and the owner who can act on the result. The initial discussion should establish which evidence is already available and which uncertainty is material enough to investigate.

The specific failure hypothesis is that content classified as a harmless example is executed with broader permissions than intended. It remains a hypothesis until examined in the deployment. Existing internal tests or operating records may already resolve it; identify the uncovered question before commissioning additional work.

Mechanism

Execution controls must govern interpreters and side effects

Relationships within runtime isolation

Source or conditionExecution approval
Define which code paths the task may invoke.
Crossing or choiceRuntime restriction
Constrain file, process, and network effects.
Consequence or decisionObserved behaviour
Use harmless simulations to inspect enforcement.
Compare effective controls with harmless test observations. Each relationship needs evidence in the actual scope.

What this specialty examines

Generated-code testing should separate source trust from execution privilege. A script copied from a document remains untrusted input even if its surrounding prose looks helpful. Identify who approves execution and which interpreter, filesystem and network capabilities are available afterward.

The review should use authorised non-destructive simulations to inspect execution approval, isolation and observable side effects. Identify the normal route, relevant exceptions and the authoritative record of the outcome. Inspect the relevant business-system record, enforcement event or process evidence alongside any reported status. Keep incomplete cases visible, with an explanation of what their absence prevents the owner from concluding.

The tradeoff to make explicit

For this decision, automated execution improves workflow speed while the trust of code sources remains a separate decision. Compare the available design or operating choices against the same intended outcome. If one option removes a required control or changes the task boundary, describe that difference before treating its apparent simplicity, speed or lower effort as an improvement. The decision should expose the tradeoff instead of burying it in an aggregate score.

Preparation

Connect the working inputs to the boundary being examined

Evidence packet for execution approval

Illustrative working format

BasisExecution policies
Confirm the applicable revision and who resolves disputed interpretation.
ContextAllowed interpreters
Check its relationship to the relevant workflow and expose missing context.
ObservationNetwork limits
Name the custodian and the conditions requiring a refreshed version.
DependencyHarmless test payloads
Identify provenance and the specific decision this material can inform.
Illustrative preparation record. Availability, revision, and handling need confirmation.

Start with execution policies, allowed interpreters, network limits and harmless test payloads. Identify the owner, revision and operating context of each source. Sanitised examples may support initial design, but differences from actual conditions should remain explicit. Where direct access is unsuitable, a customer-controlled evidence or execution route may be considered, with agreed handling and retention.

The scope also needs the runtime owner, agreed expectations for a representative boundary test, and the authority to inspect the relevant boundary. Specify what evidence would distinguish the failure in which content classified as a harmless example is executed with broader permissions than intended from acceptable behaviour. If stakeholders disagree about intended behaviour, resolve the interpretation before classifying the observed difference as a defect.

How the work proceeds

Carry execution approval through to an accountable finding

From the scoped material to a decision record

  1. Establish conditionsExecution policies

    Declare the permitted environment and resources.

  2. Examine the relationshipRuntime restriction

    Apply the scoped check to observed behaviour.

  3. Retain the findingControlled execution test plan

    Record the evidence, open dependencies, and the owner’s next decision.

Proposed sequence; resolve interpretation or access gaps before relying on the next step.

Prepare the runtime boundary and fixtures

Use execution policies, allowed interpreters, network limits and harmless test payloads to identify the exact runtime and boundary claimed by application security specialists and execution-platform owners. Mark permitted resources and protected resources in a controlled fixture, including shared services where they are relevant. Define observable success, denial and incomplete setup before execution. The preparation output is a boundary map and fixture manifest that lets reviewers distinguish effective isolation from a test that never reached its target.

Exercise and inspect the boundary

Use harmless simulations that exercise the control path without destructive payloads. Observe approval binding, command construction and actual side effects. Code inspection and runtime containment answer related but different questions and should not substitute for one another.

Choose containment or configuration changes

Translate the observations into a controlled execution test plan and findings tied to policy enforcement. Locate each issue in the worker, shared service, policy rollout or cleanup path rather than attributing every symptom to the sandbox as a whole. Decide which configuration change, narrower resource exposure or specialist investigation is warranted. Retest the affected boundary and inspect residual state before treating the changed environment as equivalent to the intended design.

Indicative artefact

Make the controlled execution test plan usable by its owner

Controlled execution test plan

Illustrative working format

Annotated fieldControlled execution test plan
Identify the assessed workflow, revision, and decision boundary.
Annotated fieldPolicy enforcement
Identify the governing revision, exception path, and accountable interpretation owner.
Annotated fieldRuntime restriction evidence
Link observations and counterevidence to the case, configuration, and conclusion limit.
Illustrative artefact structure, not a completed engagement or an observed result.

A useful decision artefact is a controlled execution test plan and findings tied to policy enforcement. Supporting material may include selected cases, configuration or process records, evidence references, a disagreement log and prioritised actions. Retain the case selection rationale and material disagreements alongside the result; otherwise the result may be mistaken for complete coverage.

Illustrative scenario

Examine observed behaviour through a constructed exception

A constructed case and its decision paths

Illustrative scenario

  1. SituationExecution approval

    A permitted script launches an interpreter outside the expected policy path.

  2. Question to examineRuntime restriction

    Trace the permitted script's interpreter and harmless side effects.

  3. Conditional next actionObserved behaviour

    Restrict or explicitly authorise the additional execution path.

Fictional example. No executed test, customer engagement, or measured outcome is implied.

Consider a fictional situation: an agent copies a script from a retrieved page and runs it without distinguishing explanation from executable instruction. The scenario is illustrative, with no client engagement, executed test or measured improvement implied.

For the retrieved script, test whether the workflow presents executable content and requested permissions clearly. A user may choose to review or reject it. The design should not convert a request to explain code into permission to run it.

Scope and next action

Keep the conclusion within the authorised runtime boundary

Confirm fit before the next commitment

Included boundaryExecution approval
This review is limited to the authorised runtime boundary.
PrerequisiteExecution policies
The relevant owner must confirm access and the intended operating conditions.
Claim limitControlled execution test plan
Untested surfaces remain outside the conclusion.
Bring the triggering change and existing evidence; resolve the dependency that could alter the decision.

Testing is confined to explicitly authorised environments and non-destructive cases. A bounded assessment cannot establish that escape or cross-boundary access is impossible. Any engagement depends on an agreed scope, qualified team, authorised access and confirmed availability. Bring the affected workflow, its current configuration or process, the triggering change and the evidence already available. The immediate scope question is prompted by the situation in which an agent is permitted to execute code obtained from untrusted task content; confirm which missing observation would change the owner’s next action.

Ways to engage

Related engagement scopes

AI Agent Runtime Security Review

Make every delegation explicit. Examine who the agent acts for, where it can execute and which tools it can reach when instructions alone are not enough.

Read about this offer

AI Agent Red Team

Challenge the boundaries that matter to your workflow. Test agreed attack scenarios against agent instructions, tools and data access.

Read about this offer

AI Agent Sandbox Assurance

Examine the boundary between agent execution and the systems around it, including files, networks, secrets and cleanup.

Read about this offer

Continue exploring

Related insights and resources

Agent security6 MIN READ

The permission boundary behind every agent action

An agent can only be evaluated in the context of what its tools allow. Examine identity, delegated authority and action boundaries before you put a workflow in motion.

Whitepaper5 MIN READ

Before your agent can act, define its authority

A working paper for the moment an AI system moves from advice to action. Map identity, permissions, approvals and recovery into an authority model your team can review and test.

A little more clarity

Questions to consider.

Start with the questions that shape a better decision.

Does scanning code remove the need for an execution boundary?

No. Scanning and runtime boundaries answer different questions. Code inspection may identify concerns, but execution still needs controlled permissions, resources and network access appropriate to the task.

Which policies authorise generated-code execution?

Start with execution policies, allowed interpreters, network limits and harmless test payloads. Record their revision, owner and relationship to the intended workflow. Missing context should be identified before interpreting outcomes; a sanitised fixture is useful only when its relevant differences from the operating case are understood.

How should untrusted scripts be tested safely?

Use authorised non-destructive simulations to inspect execution approval, isolation and observable side effects. Retain effective runtime configuration and observations from the relevant boundary. Distinguish an enforced denial from failed setup or an operation that never reached the resource. Use controlled markers and preserve residual-state evidence.

What should an execution-control report explain?

The proposed output is a controlled execution test plan and findings tied to policy enforcement. Include the selected conditions, supporting observations, unresolved differences and next decision. A reviewer should be able to follow each important conclusion back to its evidence and understand the limits without depending on an undocumented explanation from the author.

When should code remain a reviewed artefact?

Keep code as a reviewed artefact when execution is unnecessary, authority is missing or the required boundary has not been established. A request to explain or generate code does not automatically authorise running it.

What if existing evidence already answers the question?

Review its scope, provenance and configuration before proposing new work. Existing work from application security specialists and execution-platform owners may already be sufficient. In that case, capture the decision, unresolved limitations and the change that would require another review. Additional assessment should address a real gap.

What would make the findings inconclusive?

Missing execution policies, allowed interpreters, network limits and harmless test payloads can prevent a reliable interpretation, depending on the case. Record the specific missing observation and its implication. The owner can supply evidence, narrow the claim or defer the dependent decision; an incomplete case should not silently be counted as a pass.

Start with the decision you need to make.

Tell us about the agent workflow you want reviewed, and where you need greater clarity. Scope and availability are confirmed together.

Book a 30 minute readiness call Get the agent release checklist

Connected disciplines

Persistence and cleanup verification

Verify that finished agent sessions leave no unintended reusable state. The review connects session lifecycle events, volume policies, artefact lists and synthetic state markers to a bounded review of execution boundaries, giving the runtime owner a clearer basis for the next action.

Tenant-isolation assessment

Assess separation between tenants sharing agent execution infrastructure. The review connects tenant allocation rules, storage namespaces, cache keys and synthetic tenant cases to a bounded review of execution boundaries, giving the runtime owner a clearer basis for the next action.

Sandbox image supply-chain review

Control the provenance and contents of sandbox images. The review connects image manifests, build recipes, dependency inventories and signing or provenance records to a bounded review of execution boundaries, giving the runtime owner a clearer basis for the next action.