Logo
LogoDeck
  • Pricing
  1. Blog
  2. Product

Your Customer Evidence Now Lives Where You Build

Deck plugins bring customer evidence into Claude Code and Codex so product and design teams can use AI for discovery, feedback analysis, prioritization, initiative briefs, Project briefs, and NPS breakdowns.

Matt Teixeira

05 Jun 2026

4 minutes read

Deck's plugin for Codex and Claude Code & Skills

Product teams have a split problem. Customer feedback sits in one place. The work of building sits in another.

You run discovery in Deck. You write specs, prototypes, and code in Claude Code or Codex. Between the two, the evidence gets summarized, paraphrased, and slowly disconnected from the customer who said it.

And the faster the work moves, the easier it is to lose the customer context that should shape it. Agents make shipping quick. They do not make it right on their own.

Today we are closing that gap. The Deck plugin brings your customer feedback directly into Claude Code and Codex.

Chat with Your Data

The plugin connects Deck's hosted MCP server to your coding agent and installs a set of Deck-specific skills on top of it.

That means your agent can read real customer evidence: themes, insights, transcripts, segments, NPS, opportunities, initiatives, Projects, and recent feedback, without leaving the place where you are already working.

No API keys. No copy-paste. The feedback comes to the work.

For Modern, AI-native Product Teams

This is for product teams who use AI agents to build, and who do not want to lose the customer along the way.

Product managers can ground roadmap decisions and initiative briefs in evidence, not memory. Designers and researchers can pull verbatim quotes and usability patterns straight into a prototype session. Founders running lean can move from "what are customers saying" to "here is what we are shipping" in one place.

If your team has started building with agents, this keeps those agents honest. Every answer stays tied to a real customer voice.

From Product Discovery, to Prioritization, to Generating Project Evidence

The plugin ships with skills for the work product teams actually do:

SkillWhat product teams can use it for
discovery-partnerExplore customer problems, validate ideas, and decide what to investigate or build next from Deck themes, insights, segments, and customer evidence.
feedback-analystRun deeper investigations into churn, adoption blockers, feature requests, segments, source feedback, and satisfaction drivers.
initiative-brief-generatorTurn an initiative link, prompt, page, or evidence set into a concise initiative brief with customer problem, segment, scope, success metrics, risks, and evidence appendix.
project-brief-generatorTurn a Project link, Jira or Linear reference, prompt, page, or evidence set into a concise Project brief with customer problem, execution scope, success metrics, risks, and evidence appendix.
nps-breakdownUnderstand NPS by promoters, passives, detractors, segment-level scores, drivers, related opportunities, representative feedback, and movement over time.
prioritisation-advisorCompare roadmap bets using feedback volume, sentiment, segment concentration, feature requests, NPS impact, recency, and customer quotes.

Ask in plain English. The agent does the work.

For prioritization, you might ask:

Use Deck to compare the top three roadmap opportunities for activation.
Rank them by evidence strength, customer reach, sentiment, recency, NPS impact, and risk.

For Project planning, you might ask:

Use Deck to generate a Project brief for the Linear-linked onboarding cleanup work.
Include the customer problem, parent initiative, evidence, execution notes, success metrics, and risks.

Install it

Claude Code

claude plugin marketplace add getdeck-io/plugins
claude plugin install deck@deck-plugins

Then run /mcp and authenticate with your Deck account.

Codex

codex plugin marketplace add getdeck-io/plugins
codex plugin add deck@deck-plugins
codex mcp login deck

An org admin needs to enable MCP access in Deck under Settings -> MCP first. Authentication runs through browser OAuth, so there are no keys to manage.

Not using a coding agent? Use the skills in Claude Chat

You do not need Claude Code or Codex to put these workflows to work. The same skills are available as copy-paste prompts you can drop straight into Claude Chat.

Nothing gets installed. You copy a skill, paste it in, and your conversation runs the same evidence-first workflow: prioritization, idea validation, feedback investigation, NPS analysis, or initiative and Project briefs.

Here is how:

  1. Connect Deck via MCP. This is what gives Claude access to your real customer evidence. Without it, the skill has the workflow but nothing to ground it in. Set up MCP.
  2. Open the AI Skills page and pick the product conversation you need.
  3. Copy the skill prompt for that workflow.
  4. Paste it into Claude Chat as a message, or as a project instruction so every chat in that project uses it.
  5. Ask your question in plain English. Claude follows the skill and pulls the evidence from Deck.

Browse the skills here: getdeck.io/resources/ai-skills.

Start here

The plugin is most useful when there is real feedback behind it. If your team is not on Deck yet, that is the place to begin.

Sign up for Deck.

You can find the plugin repo at github.com/getdeck-io/plugins.

Footer

Logo

The AI Product Manager for Customer-Led Decisions.

Product

  • Integrations
  • Projects
  • MCP Server

Resources

  • Security
  • AI Skills
  • Documentation
  • Contact
  • Changelog
  • Story

Legal

  • Terms of Service
  • Privacy Policy

© 2026 Soho Tech Holdings Pty Ltd, All rights reserved

X (formerly Twitter)LinkedInYouTube