Docs

MindForge Guard docs

Use these docs to understand the deterministic governance evidence layer, the first single-agent workflow, and the commercial boundaries that keep Guard recommendation-only and human-review-oriented.

Documentation paths

Product site

Use mindforge.run for the primary commercial story: Make AI-assisted work reviewable before it becomes trusted.

mindforge.run

Repository docs

Use the public repository for workflow guides, install notes, commercial boundaries, and verification materials.

Repo docs

License Hub

Use License Hub after purchase to view licenses, delivery visibility, and account surfaces without changing Guard runtime posture.

After purchase: generate your first governance report

Use the same purchase email to sign in, download your signed license, install Guard locally, and generate a deterministic governance report for human review.

1. Sign in with your purchase email

Use the same purchase email from checkout when you sign in to License Hub.

2. Download your signed license JSON

Download the signed license JSON delivered through License Hub.

3. Install Guard

npm install -g @veeduzyl/mindforge-guard@7.0.1

4. Verify and install the license

guard license verify --file downloaded-license.json
guard license install --file downloaded-license.json
guard license status

5. Generate your first governance report

guard pack validate --pack examples/single-agent-governance-pack/hr-self-service-agent --preview --json
guard report single-agent --pack examples/single-agent-governance-pack/hr-self-service-agent --preview --json

6. Read the report

authority boundary, execution evidence, missing evidence, risk/drift signals

7. Make the next human review decision outside Guard

The final human review decision remains outside Guard.

Guard produces review evidence. It does not approve, block, deploy, certify, or control execution.

Single-agent governance report docs

An Evidence Pack is the review bundle behind an AI-assisted action: task context, allowed scope, action trace, tool/data references, outputs, missing evidence, and reviewer notes.

These docs explain how Guard turns that evidence bundle into a governance report so reviewers can inspect authority boundary, execution evidence, missing evidence, and risk/drift signals.

Single-agent workflow guide

Move from a sample agent action to a governance report that surfaces authority boundary, execution evidence, missing evidence, and risk/drift signals.

Open doc

Single-agent positioning brief

Read the commercial boundary freeze for the deterministic governance evidence layer story.

Open doc

First governance report guide

Use the current public guide for install, first-run validation, report generation, and human review reading flow.

Open doc

What Guard does not do

See the recommendation-only, non-executing, non-control-plane boundary in buyer-readable language.

Open doc

Secondary technical install

For local validation, install the current recommended v7.0.1 package: npm install -g @veeduzyl/mindforge-guard@7.0.1.

Then use the workflow guides on this page to generate your first governance report.

Trust and workflow pack

This is the canonical trust and workflow entry for License Hub visitors evaluating Guard before they choose a license.

First 10 Minutes With Guard

Start with the current first-run path before you choose a paid license.

Open guide

Safety Boundary

Understand what Guard does and does not do before it enters higher-risk delivery paths.

Open guide

Choose the Right Guard Edition

Review the buyer-readable edition value map before upgrading.

Open guide

Current Product Demos

Run the current demos before upgrading or choosing a team rollout path.

Open guide

Trust FAQ

Use the current FAQ for trust, positioning, and commercial boundary questions.

Open guide