Runtime Safety

Autonomy needs an independent safety boundary.

AI agents inherit the operator’s shell, filesystem and credentials. CodeMarine evaluates defined high-impact actions before execution on supported paths, then shows where that boundary ends.

Runtime evaluation is implemented. Proof-backed Guarded status, live canaries and the unified Protection Center are planned.
Runtime Protection Current product UI · demo workspace
CodeMarine Runtime Protection view showing the coverage status of supported development agents
Captured from CodeMarine desktop with local demo data.
The trust problem

Permission prompts ask the wrong question.

A broad prompt asks whether an agent may use the shell. Runtime policy asks whether this exact operation matches a deterministic catastrophe rule. Those are different decisions.

PROVIDER PERMISSIONMay the agent run shell commands?

Useful for setting a broad authority boundary. Repeated prompts can still create fatigue.

CODEMARINE POLICYIs this proposed operation catastrophic?

A narrow, independent decision for a defined action on an observed surface.

Deterministic rule packs

Catastrophe has recognizable shapes.

The current evaluator covers nine operational rule packs. The UI should group them by the outcome a developer understands, not by implementation file.

01

Files and permissions

Recursive deletion, protected boundaries, permission changes and disk operations

02

Git history

Destructive clean, hard reset, branch deletion and force-push signals

03

Databases

Schema drops, object deletion, truncation and broad destructive mutation

04

Containers and Kubernetes

Pruning, volume removal, namespace deletion and broad resource removal

05

Cloud and infrastructure

High-impact resource deletion across cloud CLIs, Terraform and related tooling

06

Credentials and integrity

Secret source-to-sink signals, protection tampering and sensitive configuration access

Structured decision Target architecture
TRUSTED CONTEXT
Operation
Delete resource
Environment
Production
Blast radius
Broad
Reversible
No checkpoint
Privilege
Elevated
POLICY RESULTBLOCK

Immutable catastrophe policy. The agent cannot approve its own exception.

Beyond command text

Reason about the operation and its authority.

A command string is only one signal. Strong production policy also needs trusted environment identity, resolved target scope, reversibility and the credentials being used.

Immutable blocksDisk boundaries, protection tampering and clear credential exfiltration
Production blocksDestructive data or infrastructure operations where target context is independently proven
Review requiredHigh-impact actions that need a safe human rerun or future signed approval
Warn or observeAmbiguous or reversible operations below the blocking threshold
Protection depth

The control must sit where the authority flows.

A laptop hook cannot govern a remote agent. A shell parser cannot see a native SDK call. CodeMarine expands coverage by placing policy at several real interception points.

Beta 01

Local host hooks

Evaluate supported shell and write events before the provider executes them.

Planned 02

MCP and API gateway

Inspect structured tool calls before forwarding them to databases, SaaS systems or internal APIs.

Planned 03

Scoped credentials

Give the agent short-lived authority for one approved operation instead of a standing production secret.

Planned 04

Remote and downstream policy

Use sidecars where possible, then reinforce the boundary through CI, SCM and cloud controls.

Protection Center

Make safety obvious without making the UI noisy.

The planned experience elevates Runtime Safety into the normal product view. It separates code findings, action prevention, after-effect detection and recovery.

Sound is an attention channel. It is not evidence that an action was blocked. Only a proof-backed host confirmation can support that claim.

Per-surface statusGuarded, Alerts only, Setup needed, Problem, Off or Not covered

Critical incidentPlain-language reason, agent attribution and a safe next step

Native notificationImmediate host feedback without exposing the raw command

Optional soundUser-controlled, rate-limited and reserved for eligible critical events

Coverage contract

Prevention, detection and recovery are different.

A file watcher can detect a resulting workspace change after it happens. It cannot prove that a database call or cloud deletion was prevented.

CapabilityMeaningCurrent status
PreventionAn authoritative hook denies the action before execution. Beta
DetectionGuardian observes an eligible resulting file change while the watcher is healthy. Current
RecoveryA prior captured file state can be restored after mutation. Planned
Native or remote actionAn unmediated SDK, browser, desktop or hosted-agent action. Not covered
Runtime Safety evaluation

Test the boundary before you trust it.

CodeMarine is building proof-backed protection for teams that want agent autonomy without hiding the gaps.

Sarge, the CodeMarine guardian