Category Deep Dive · Infrastructure for the Age of Agents

Cloud Infrastructure Built for AI Coding Agents

Agents are the new infrastructure consumers. Hand them thirty cloud services and they inherit your cloud. Hand them one sandboxed environment with a ceiling and they ship production-ready systems safely.

Agent Sandboxesstack.json IaCHosted GitCollaboration BoardApproval Workflows
Talk to us Read Technical Documentation → Software Lifecycle Cloud →

Why coding agents need a lifecycle-native runtime

Modern coding agents (Cursor, Claude Code, Codex, Pi) can write multi-file features in seconds, but they hit a wall at the boundary between code and cloud. When infrastructure is defined across imperative console clicks, scattered Terraform modules, and external dashboard secrets, an agent cannot safely provision a new table, test a webhook, or verify a search index.

KERNSTACK makes the entire cloud topology legible to agents in a single stack.json file while enforcing a hard safety boundary around production.

  • Deterministic schema: agents read stack.json and .kernstack/sdk--v1 without hallucinating cloud APIs
  • Isolated sandboxes: every agent run operates on an agent/ branch and isolated preview environment
  • Explicit ceiling: org administrators cap agent credit spend and require human approval before production deploy

The KERNSTACK Cursor plugin and CLI preflight workflow

Whether an engineer drives an agent locally through the KERNSTACK Cursor plugin or a stakeholder mentions @agent in a Telegram thread, the workflow follows the same governed discipline: auth preflight, kernstack plan resource diffing, isolated preview deployment, and human sign-off.

Pluggable harnesses and European model options

Intelligence improves every quarter. KERNSTACK supports Cursor (Composer, Grok), Pi (Devstral, Mistral Large/Medium), Claude Code, and Codex against the exact same project graph — so upgrading your agent never requires replatforming your cloud.

FAQ

Frequently asked questions

No. Agents execute inside isolated sandboxes bound to preview environments with a strict credit ceiling. Production environments and main branches require human engineer approval.

The CLI vendors strongly typed SDKs directly into .kernstack/ in the repository, and the entire infrastructure schema is declared in stack.json.

Yes. Partners or product managers can ask @agent for a change in the project conversation; the agent deploys to an isolated branch and posts a preview link without touching main.

Connected in the Project Graph

Related capabilities, architectures, and comparisons

Category PillarThe Software Lifecycle CloudThe complete architecture guide to vertically integrated cloud infrastructure for people and agents.Category Deep DiveSoftware Lifecycle Cloud vs Hyperscalers (AWS, GCP, Azure)Why European engineering teams choose KERNSTACK over assembling 15+ AWS, Google Cloud, or Azure services: one control plane, one stack.json, zero US lock-in.Category Deep DiveSoftware Lifecycle Cloud vs Internal Developer Platform (IDP)Why 50–250 person companies choose KERNSTACK, the Software Lifecycle Cloud, instead of spending 18 months building an Internal Developer Platform on AWS.Category Deep DiveSoftware Lifecycle Cloud vs Traditional PaaSCompare the Software Lifecycle Cloud against traditional PaaS providers like Vercel, Heroku, Render, and Railway across data, AI, agents, and EU bare metal.Category Deep DiveWhat Is the Compounding Project Graph?Discover how KERNSTACK connects conversations, tasks, git branches, stack.json infrastructure, runtime telemetry, and analytics into one compounding projectPlatform · AgentsAgent sandboxesIsolated sandboxes for coding agentsPlatform · PeopleCollaborationBoards and conversations on the projectPlatform · SourceHosted GitRepository hosting inside the projectPlatform · Manifeststack.jsonOne manifest for the whole projectPlatform · OrchestrationWorkflowsDurable workflows with human pausesUse Case · AgentsAI agent backendHosted git carries the repository; agent branches namespace the experiments. The agent edits code and stack.json together, runs kernstack plan --json to verify its own change set, and a reviewer checks the same plan. Environments multiply cheaply because a project is declared, not hand-assembled: the agent-<id> environment runs the branch’s stack for click-through review.Use Case · OperationsInternal operations toolStart from the spreadsheet’s columns: they become a table keyed by the request’s natural id, with status, owner, and timestamps as fields. A minimal app — your framework under apps/, built and published by deploy — reads and writes through functions; files attach from buckets; and the workflow layer carries approvals as human steps, so the state “waiting on finance” is a place in a flow, not a cell someone coloured yellow.Use Case · SupportRAG knowledge assistantDocuments land in a bucket; a vector index is declared over that bucket and re-embeds from source when articles change. A Go function behind a gateway route embeds the question, retrieves the closest passages, and calls a catalogue or custom inference endpoint with those passages as context. Query logs go to a table, so unanswered questions become an editorial list.

Ship your stack on the Software Lifecycle Cloud.

Closed preview on dedicated European infrastructure. Bring a repository or start from a blueprint.

Talk to us See pricing & credits →