Skip to content
Scan one machine free. No account. Nothing leaves your machine.
KeepRails

Understand the category

What is AI-tooling evidence?

It is evidence about the tools configured around an AI coding agent: what is installed, where it came from, what scope it has, what permissions it declares, and what can be concluded from that setup without collecting prompts or code.

The category

Endpoint Shadow AI system of record

The system of record for endpoint-resident Shadow AI: a data-local inventory and change history for AI apps, agents, models, IDE tools, MCP servers, browser extensions, SDKs, and provider-key presence across the fleet.

Inventory

MCP servers, skills, plugins, their user or project scope, provenance, version, transport, and declared permissions.

Evidence

A finding names the configuration fact and rule behind it instead of collapsing the setup into a score.

Change control

A deterministic fix is previewed, approved, applied with a backup, verified, and reversible.

How it works today

Begin with facts one machine can prove

The current product does not need a dashboard, gateway, or device-management system to produce its first result.

  1. 01

    Discover the setup

    Read local configuration and resolve which assets are present at which scope.

  2. 02

    Explain findings

    Report duplicate definitions, scope conflicts, unverifiable provenance, version drift, and explicit risk with evidence.

  3. 03

    Offer a reversible path

    Show the exact change first, wait for approval, keep a backup, and verify the result.

Boundaries

What the category is not

Not a prompt monitor

Prompts, responses, source code, diffs, shell text, and tool payloads are outside the data contract.

Not a gateway

KeepRails does not sit in the tool-call path and does not claim runtime blocking.

Not employee scoring

No developer score, ranking, leaderboard, or causal productivity claim belongs in this category.