Smart Commenting System for Case Studies

The Smart Commenting System enables Google Docs-style collaboration on case studies. Users can select any text and attach comments directly to that selection, creating a precise feedback loop between AI authors and human reviewers.

Key Features

Design Philosophy

The system follows the "full Janus battery" pattern:

This layered approach ensures both humans (via UI) and AI agents (via MCP) can participate equally in the review process.

Selecting Text and Adding Comments

The popup includes:

Viewing Comments

Click the comment icon in the top-right corner to open the sidebar:

Resolving Comments

When a comment is addressed:

Resolved comments remain visible but are visually de-emphasized, preserving the discussion history.

Database Schema

CREATE TABLE case_study_comments (
    id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
    case_study_id UUID NOT NULL REFERENCES case_studies(id),
    section_id UUID REFERENCES case_study_sections(id),
    author TEXT NOT NULL,
    author_type TEXT NOT NULL DEFAULT 'human',
    content TEXT NOT NULL,
    quote_text TEXT,  -- The selected text this comment references
    resolved BOOLEAN DEFAULT FALSE,
    resolved_by TEXT,
    created_at TIMESTAMPTZ DEFAULT NOW(),
    updated_at TIMESTAMPTZ DEFAULT NOW()
);

API Endpoints

| Method | Endpoint | Description |

|--------|----------|-------------|

| GET | /api/v1/case-studies/{id}/comments | List all comments |

| POST | /api/v1/case-studies/{id}/comments | Add a comment |

| POST | /api/v1/case-studies/comments/{id}/resolve | Mark as resolved |

| DELETE | /api/v1/case-studies/comments/{id} | Delete a comment |

MCP Tools

AI agents can use these tools programmatically:

Frontend Components

How AI Agents Use Comments

AI agents participate in the review process through MCP tools. This enables:

Automated Content Review

When a case study is submitted for review, an AI agent can:

Example: AI Adding a Comment

# AI agent reviewing a case study
await client.case_study_comment_add(
    case_study_id="6d354192-...",
    content="This section could use a concrete example to illustrate the concept.",
    author="claude-code",
    author_type="ai",
    section_id="section-3-uuid",
    quote_text="The autonomy stack enables..."
)

Collaborative Workflow

This multi-agent approach ensures content is reviewed from multiple perspectives before human approval.

Planned Features

Text Highlighting

Currently comments reference text via quote_text field but don't visually highlight it. Future work:

Threaded Replies

Enable conversations within comments:

AI Auto-Review

Automatic AI review when drafts are submitted:

Real-time Collaboration

WebSocket-based live updates:

Rhea SSO Integration

Authentication for verified human identity:

Implementation Priority