Sandbox Policy Design with Claude Code: API Runtime Pattern
A production playbook for sandbox policy design in cross-industry operations using Claude Code: api runtime pattern, run-scoped inputs, logs, typed results, and artifacts.
Audience: Security reviewers and platform teams
The problem
Security reviewers and platform teams need sandbox policy design to run repeatedly against tool policies, file scopes, and network rules. In cross-industry operations, the pain is not one good answer; it is repeatability, auditability, exception handling, and evidence that survives handoff.
Implementation path
Package the sandbox policy design instructions as a skill, send tool policies, file scopes, and network rules as run-scoped inputs, execute with Claude Code, poll terminal status, and consume argo.result.v1 instead of parsing a transcript.
Tradeoffs and failure modes
The API boundary forces the workflow to define inputs, terminal states, and result shape before customers depend on it. For sandbox policy design, the practical test is whether a second run can be debugged, retried, and consumed by a product without reading the raw agent transcript.
Run request
POST /api/skills/<skill_id>/run
provider=claude-code
workflow=sandbox-policy-design
inputs[]=@./input-pack.zip
result_schema=argo.result.v1
Run this on Argo