Ddx

DDX Plugin for Claude Code

MCP plugin that connects Claude Code and Claude Desktop to MedScout's DDX platform — Voice of Customer (VOC) and Go-to-Market (GTM) account research.

Setup

1. Get Your API Token

1. Log into DDX at https://ddx.medscout.io
2. Visit /settings/mcp/ — your token is auto-created
3. Copy the token

2. Set the Token

Add to your shell profile (~/.zshrc or ~/.bashrc):

export DDX_API_TOKEN="<your-token>"

Then reload your shell (source ~/.zshrc) or open a new terminal.

3. Connect

Claude Code:

/plugin marketplace add medscout-dev/ddx-claude-plugin
/plugin install ddx@medscout

Claude Desktop: Copy the JSON config from the setup page into your claude_desktop_config.json.

What You Get

Feature	Claude Code	Claude Desktop
21 MCP tools	Yes	Yes
7 resources	Yes	Yes
5 prompts	Yes	Yes
7 skills (/submit-voc, etc.)	Yes	No
VOC Analyst agent	Yes	No
GTM Research Analyst agent	Yes	No

Available Skills

Skills are guided workflows you trigger with slash commands in Claude Code.

VOC Skills

Skill	What it Does
/submit-voc	Gathers details, searches for duplicates, then creates or matches to an existing VOC item
/analyze-feedback	Paste meeting notes, transcripts, or emails — extracts structured VOC items with duplicate matching
/voc-search	Search items, browse themes, analyze trends, query transcripts, view company feedback history
/merge-voc	Merge two VOC items — AI-suggested content, target recommendation, full relationship transfer

Platform Skills

Skill	What it Does
/feedback	Submit feedback or a bug report about DDX directly to the #feedback-ddx Slack channel

GTM Skills

Skill	What it Does
/research-search	Search research packs, read report content, check request status
/generate-research	End-to-end research generation: transcripts → reports → published pack → completed request

Available Tools

VOC Management

• search_voc_items — Keyword search across title, description, and problem statement
• find_similar_voc_items — Semantic similarity search via AI embeddings
• get_voc_item — Full item details with companies, themes, tags, and comments
• create_voc_item — Create a new VOC item (auto-generates embeddings)
• add_company_to_voc_item — Associate a company with priority and context
• add_attachment_to_voc_item — Upload a file (image, PDF, etc.) to a VOC item from a URL
• search_companies — Find companies by name, filter by health status

Analytics

• list_themes — Browse themes with business metrics (ARR, company count, health)
• get_theme_details — Deep dive into a theme with items, companies, sub-themes
• get_voc_analytics — Flexible analytics: group by theme/company/status/tag, compare periods, semantic filter

Transcripts

• query_transcripts — Filter transcripts by company, team, date; keyword search across full text
• search_transcripts_semantic — Find transcripts by meaning using AI embeddings

Bulk Processing

• parse_feedback — Parse unstructured text into structured VOC proposals with automatic deduplication

Platform Feedback

• submit_platform_feedback — Post feedback or bug reports about DDX to #feedback-ddx Slack channel

VOC Merge

• suggest_merge — AI-suggested merge content with target recommendation and relationship preview
• execute_merge — Execute the merge with full relationship transfer (companies, themes, tags, features, attachments, comments, mentions)

GTM Research

• list_research_packs — List research packs with company and report summaries
• get_research_report — Get full report content (fingerprint, AE briefing, CS briefing, full pack)
• publish_research_pack — Create or update a research pack with reports
• list_research_requests — List pending/submitted research requests
• complete_research_request — Mark a request as completed and link to a pack

Agents

VOC Analyst

Autonomous agent that answers product strategy questions using real customer data. Automatically selects the right tools, gathers data across multiple dimensions, and synthesizes findings into actionable insights with specific numbers (ARR, company count, health status).

Triggered automatically when you ask broad product questions like:

• "What are customers most frustrated about?"
• "Which themes have the highest at-risk ARR?"
• "Compare CS vs Sales feedback themes"

GTM Research Analyst

Autonomous agent that generates account intelligence from call transcripts and company data. Processes research requests end-to-end: gathers transcript intelligence, generates reports, publishes packs, and completes requests with Slack notifications.

Triggered automatically when you ask about account research:

• "Generate research for Acme Corp"
• "What research requests are pending?"
• "Show me the latest research packs"
• "Process the next pending research request"

Example Queries

You can ask questions in natural language — no slash commands required:

"What are the top VOC themes by ARR?"
"Search for items about profile usability"
"What has Acme Corp requested?"
"Show me the fingerprint for acme-corp"
"What research requests are pending?"
"Generate research for company XYZ"
"What calls mentioned data exports this month?"
"Parse this meeting notes into VOC items: ..."
"Merge VOC items 123 and 456"

How It Works

The plugin connects to DDX's MCP server at https://ddx.medscout.io/mcp using Streamable HTTP transport. Your DDX_API_TOKEN is sent as a Bearer token in the Authorization header. The MCP server runs inside Django, accessing the same database as the DDX web app — so data is always live and consistent.

Troubleshooting

"Missing or malformed Authorization header" (401)

• Check that DDX_API_TOKEN is set: echo $DDX_API_TOKEN
• Reload your shell after adding it to .zshrc/.bashrc

"Invalid or inactive API token" (401)

• Your token may have been regenerated. Visit /settings/mcp/ to get your current token.

Plugin not finding tools

• Verify the plugin is installed: claude plugin list
• Reinstall if needed: claude plugin remove ddx then /plugin install ddx@medscout

CI/scripting environments (no browser)

• For headless environments, use a static API token instead:

  claude mcp add --transport http ddx https://ddx.medscout.io/mcp \
    --header "Authorization: Bearer <your-token>"
  

• Get your token from /settings/mcp/

Architecture

plugin/
├── .claude-plugin/
│   └── plugin.json         # Plugin metadata (name, version, author)
├── .mcp.json               # MCP server connection config (HTTP URL + auth)
├── skills/
│   ├── submit-voc/         # /submit-voc skill
│   ├── analyze-feedback/   # /analyze-feedback skill
│   ├── voc-search/         # /voc-search skill
│   ├── merge-voc/          # /merge-voc skill
│   ├── feedback/           # /feedback skill
│   ├── research-search/    # /research-search skill
│   └── generate-research/  # /generate-research skill
├── agents/
│   ├── voc-analyst.md      # VOC Analyst autonomous agent
│   └── gtm-research-analyst.md  # GTM Research Analyst autonomous agent
└── README.md               # This file

The .mcp.json points to the production MCP server. The ${DDX_API_TOKEN} variable is expanded by Claude Code at runtime from your environment.