Add Collaborative Agent Review Workflow with Velt

Summary

Add a collaborative agent workflow using Velt where AI agents and human reviewers can work together through contextual comments, presence, and approval actions.

Proposed location

advance_ai_agents/velt_collaborative_review_workflow/

What to build

Create a runnable enterprise-style review workflow where an agent analyzes a document, proposal, design brief, or policy draft and leaves contextual findings for humans to review.

Use Velt to show collaboration features such as agent presence, anchored comments, mentions, resolve states, and approve/reject actions. The example should demonstrate how an AI agent can participate in the same review surface as human users instead of only producing a separate report. Follow the repository's existing Nebius Token Factory configuration pattern for model calls.

Acceptance criteria

• Includes a README.md based on .github/README_TEMPLATE.md.
• Uses Velt for collaborative comments, presence, or approval workflow primitives.
• Includes a sample review surface such as a document, proposal, policy, or design brief.
• Agent findings are anchored to specific sections, claims, or UI elements.
• Human reviewers can approve, reject, resolve, or comment on agent findings.
• Includes mock/sample data so the workflow can be understood without enterprise accounts.
• Includes .env.example with no secrets and configurable model/provider settings.
• Adds one catalog entry to the root README.md under Advanced AI Agents.

Contribution notes

Please keep this to one self-contained project and one pull request. Link the pull request with Closes #<issue-number>.