PickaxeProductize OpenCode with one harness layer around the coding runtime.
OpenCode can own the coding loop. Driftstone owns the customer-facing system around it: workspace continuity, sandbox boundaries, rollout control, scoped secrets, and usage rules that keep coding sessions governable at product scale.
OpenCode
Open-source coding agent
Runtime
Give OpenCode a repeatable customer-facing harness instead of shipping raw coding sessions directly.
Harness
Driftstone centralizes customer access, rollout, usage, and scoped runtime permissions around the agent.
Outcome
Ship the runtime as a product without rebuilding the operating layer around every customer deployment.
OpenCode fits teams that want to ship coding-agent capability without exposing the raw runtime directly.
Coding agents touch repositories, secrets, packages, and long-lived workspaces. The harness layer keeps those concerns centralized so the runtime can stay focused on code execution and editing workflows.
- Keep coding-agent behavior in OpenCode while centralizing customer delivery.
- Reconnect the right workspace and sandbox state for each user session.
- Separate plan, access, and rollout rules from coding runtime logic.
Workspace continuity
Users return to the right repo context and installed environment instead of starting cold each time.
Runtime isolation
Customer sandboxes stay separated even when the underlying coding workflow is similar.
Coding agents need stronger control points once they leave an internal team and face real customers.
OpenCode can remain the coding runtime while Driftstone centralizes who can launch sessions, what environments they can reach, how usage is metered, and how runtime updates roll forward without breaking active workspaces.
- Track spend and session usage across coding workloads and model providers.
- Scope repo access, credentials, and execution permissions to the right customer boundary.
- Roll out environment updates without flattening workspace state.
Session governance
Apply plan limits, cutoffs, and access rules before coding sessions sprawl into unmanaged cost.
Scoped execution
Treat repo credentials and runtime permissions as explicit controls, not static defaults.
Controlled releases
Ship coding-runtime updates through a release path with rollback options and targeting.
OpenCode becomes much easier to support when customer delivery and runtime operations share the same control plane.
The harness layer helps teams understand where a session ran, what version it used, which credentials it received, and how customer-facing limits were applied. That makes support and growth materially easier.
- Customer support can reason about coding sessions without digging through hosts manually.
- New coding-agent deployments follow the same operational pattern.
- Billing, rollout, and access changes stop living in separate systems.
Support visibility
Session-level context makes coding incidents and customer questions easier to explain.
Scalable launch pattern
The same harness can support more workspaces and customers without a rewrite.
Runtime page
Keep OpenCode focused on code. Centralize the rest.
A coding agent can be powerful on its own. The product gets durable when workspace continuity, rollout, access, and billing become one system around it.