Overview of Pendo AI skills

Last updated: October 21, 2026 23:29

Pendo AI skills are prebuilt workflows you run inside an AI client or AI coding assistant, such as Claude, Cursor, or Codex, to turn something you find in Pendo into a concrete next action, or to set up Pendo in your codebase, without leaving the tool you already work in.

Some skills connect to your Pendo data through the Pendo MCP server. Others run in your coding assistant and work only on your codebase, so they don't need a connection to Pendo. Each skill produces a specific output, such as a health report for an app or account, an evidence-backed fix brief for your AI agent, a draft guide or Orchestrate journey in Pendo, or automated event tracking in your codebase.

These skills are separate from the skills in Leo, Pendo's built-in AI assistant, which are guided prompts you run inside Pendo. For more information, see Analyze product usage with Leo (beta).

Before you begin

Requirements vary by skill. Skills that work with your Pendo data need an AI client that supports skills, such as Claude or Cursor, and a connection to your Pendo data through the Pendo MCP server. Most of these skills are packaged as Pendo plugins for Claude Code, which install the skill and configure the Pendo MCP server connection for you. For more information, see Install the Pendo plugins for Claude Code.

Skills that create something in Pendo, such as a draft guide or a draft Orchestrate journey, use the Pendo MCP server's write tools. A subscription admin must turn on write tools in Settings > Subscription settings > AI access before these skills can save anything to Pendo. For more information, see Overview of the Pendo MCP server.

Skills that work on your codebase, such as the Agent Analytics setup, Pendo Analytics tagging, and Mobile SDK install skills, instead run inside an AI coding assistant, such as Claude Code, Cursor, or Codex, and are added to your project. For the exact requirements, see each skill's article or the plugin README.

Install the Pendo plugins for Claude Code

Pendo publishes its Claude Code skills as plugins in the pendo-plugins marketplace. Each plugin bundles one or more skills and, where a skill needs your Pendo data, configures the Pendo MCP server connection for you. You install a plugin once, then run its skills by name.

The marketplace includes the following plugins:

pendo-analytics. Account health, app health, feature adoption, feedback analysis, replay triage, and data-informed planning. Configures the Pendo MCP server and the Novus MCP server.
pendo-guides. Guide creation from a short intake conversation. Configures the Pendo MCP server.
pendo-orchestrate. Orchestrate journey creation and configuration. Configures the Pendo MCP server.
setup-agent-analytics. Agent Analytics instrumentation for the AI agents in your codebase. Doesn't need a Pendo connection.
setup-mcp-agent-analytics. Agent Analytics instrumentation for an MCP server. Doesn't need a Pendo connection. MCP analytics is in closed beta.

To install a plugin:

In Claude Code, run /plugin marketplace add pendo-io/claude-pendo-plugin.
Run /plugin, go to Marketplaces, select pendo-plugins, and install the plugins you want.
If the plugin uses your Pendo data, run /mcp and follow the sign-in flow to authenticate the Pendo MCP server. The pendo-analytics plugin also asks you to authenticate the Novus MCP server, which uses the same Pendo sign-in.
Run a skill by its full name, such as /pendo-analytics:app-health followed by the app name.

To install from a clone of the repository instead, or to see the full list of MCP tools each plugin uses, see the pendo-plugins README on GitHub.

Available skills

Select a skill to see its full requirements, invocation method, and setup steps. Skills are grouped by whether they work with your Pendo data or only on your codebase.

Skills that work with your Pendo data

These skills run through the Pendo MCP server. Unless noted, each is part of a plugin in the pendo-plugins marketplace.

Account health. Prepare for a customer call with a summary of one account's engagement, feature usage, feedback, and NPS. The skill confirms which account you mean before it reports, so partial names are fine. It's part of the pendo-analytics plugin. Run /pendo-analytics:account-health followed by the account name.
Agent repair. Turn a flagged issue in Agent Analytics into an evidence-backed fix brief that you can apply yourself, hand to a coding agent, or turn into a ticket for an engineer. For more information, see Diagnose and resolve AI agent issues with the Agent Repair skill.
App health. Generate a shareable health report for one application that covers engagement, acquisition, retention, average session time, stickiness, frustration signals, and survey scores, with a verdict and next steps. It's part of the pendo-analytics plugin. Run /pendo-analytics:app-health followed by the app name. The same report is available in Leo as the App health skill.
Data-informed planning. Ground a product plan in real usage before you write it. The skill gathers signals, issues, the product wiki, customer feedback, PES, NPS, segments, and session replays from Novus and Pendo, summarizes them as a data context block, and then writes the plan or hands it to a planning skill you already have installed. It needs both the Pendo MCP server and the Novus MCP server, and it's part of the pendo-analytics plugin. Run /pendo-analytics:data-informed-planning followed by your planning question. Novus is in open beta.
Feature adoption. Analyze how one Feature is being adopted: unique visitors and accounts, adoption trends over time, who the power users are, and which accounts haven't adopted it yet. It's part of the pendo-analytics plugin. Run /pendo-analytics:feature-adoption followed by the Feature name.
Feedback analysis. Analyze customer feedback from Listen to find themes, extract insights, and surface churn and frustration risks. You can narrow the analysis by topic, account, feedback type, alert, or date range. It's part of the pendo-analytics plugin. Run /pendo-analytics:feedback-analysis.
Guide creator. Build an in-app guide from a short intake conversation about its type (walkthrough, announcement, alert, poll, or promotion), goal, and context. The skill can match a theme from your subscription and produces a standalone HTML, CSS, and JavaScript guide file. If the Pendo MCP server is connected and write tools are turned on, it can also send the guide to Pendo as a draft. It's part of the pendo-guides plugin. Run /pendo-guides:pendo-guide-creator.
Guide management. Review the guide signals Pendo generates each week, such as guides it recommends archiving or deactivating and unusual changes in guide volume, with the evidence behind each one. You decide what happens to each guide; the skill doesn't change a guide's state unless you select the action. It's part of the pendo-guides plugin. Run /pendo-guides:guide-management. Guide signals are in closed beta.
Orchestrate journeys. Create, configure, and edit draft Orchestrate email journeys, including multi-email sequences, conditional splits, and the HTML email content, and set the audience, schedule, and goal. Activating a journey stays in Orchestrate, so nothing is sent to visitors from the skill. It's part of the pendo-orchestrate plugin. Run /pendo-orchestrate:orchestrate-journeys.
Replay triage. Move from a bug a visitor experienced in a replay to a proposed code fix in your AI client, using the console and network logs captured during the session. You can start from a replay link or from a ticket that links to one. It's also part of the pendo-analytics plugin as /pendo-analytics:replay-triage. For more information, see Use the Replay Triage skill.

Skills that work on your codebase

These skills run inside an AI coding assistant and don't need a connection to Pendo.

Agent Analytics setup. Detect the AI agents in your codebase and instrument them so prompts, responses, and visitor reactions flow to Agent Analytics. The skill reads your codebase and chooses the best of four paths: the Python SDK, the TypeScript SDK, client-side trackAgent() calls, or the server-side Conversations API. It's available as the setup-agent-analytics plugin for Claude Code and as a skill for Cursor. For more information, see Automate Conversations API implementation with an AI skill.
MCP Agent Analytics setup. Detect the language of your MCP server (Python, TypeScript, Go, or Java) and instrument it with the matching Pendo SDK so MCP tool calls flow into Agent Analytics. It's available as the setup-mcp-agent-analytics plugin for Claude Code. Run /setup-mcp-agent-analytics:setup-mcp-agent-analytics. MCP analytics is in closed beta.
Mobile SDK install. Install the Pendo Mobile SDK in your app, check an existing installation, or fix the problems you choose, without leaving your coding assistant. The install-pendo-mobile skill covers native iOS, native Android, React Native, Expo, Flutter, and .NET MAUI apps, and runs in Claude Code, Cursor, or Codex. For more information, see Developer's guide to installing the Pendo Mobile SDK.
Pendo Analytics tagging. Audit your React codebase and add pendo-analytics attributes to interactive elements, so AI tagging can suggest tags with stable selectors that are less likely to break when your code changes. It's built for engineers using Claude Code or another AI coding tool. To download the skill, see Tag with AI.

Next steps

To connect an AI client to your Pendo data, see Overview of the Pendo MCP server and Connect to the Pendo MCP server.

To set up and run a specific skill, follow the full requirements and steps in that skill's article, linked in the Available skills section, or in the pendo-plugins README.