---
title: Agent Skills
---

# Agent Skills

## Overview

bitdrift provides [skills for your coding agents](https://github.com/bitdriftlabs/bd-skills){:target="_blank"}: AI-optimized instructions that teach your agent how to use the bitdrift platform. Once installed, your agent can install and instrument the Capture SDK, create and deploy workflows, query sessions, triage issues, and more, all without leaving your editor.

!!! info
    bitdrift skills follow the [agentskills.io](https://agentskills.io/){:target="_blank"} open standard, making them compatible with any AI coding agent that supports skills, such as Claude Code, Cursor, Codex, and Copilot. They are hosted in the [bitdriftlabs/bd-skills](https://github.com/bitdriftlabs/bd-skills){:target="_blank"} GitHub repository.

## Getting started

We recommend installing bitdrift skills via [skills.sh](https://skills.sh/){:target="_blank"} cli by running the following command:

```bash
npx skills add bitdriftlabs/bd-skills
```

!!! note
    This uses `npx` for ease of installation which [requires `node`/`npm`](https://docs.npmjs.com/downloading-and-installing-node-js-and-npm){:target="_blank"}. Install it via homebrew: `brew install node`

### Updating skills

```bash
npx skills update --all
```

For other installation options and further setup instructions, see the `README` at [https://github.com/bitdriftlabs/bd-skills](https://github.com/bitdriftlabs/bd-skills){:target="_blank"}.

## Available skills

### bd-cli

Enables full use of the [`bd` CLI](../../cli/quickstart.md) from within your agent session — creating and deploying workflows, querying charts and session timelines, triaging issues, managing API and SDK keys, and more.

### bd-instrumentation

Guides integration of the [Capture SDK](../../sdk/quickstart.md) into iOS, Android, and React Native apps. The skill detects your platform, checks whether the SDK is already installed, and then either performs a fresh installation or extends an existing integration with screen tracking, network monitoring, log forwarding, custom fields, and spans.

### bd-docs

Searches and interprets this same [bitdrift documentation](../../index.md) to answer questions about product behavior, SDK setup, API references, and best practices. It fetches live docs at query time so your agent always works from up-to-date content.

## What your agent can do

The skills rely heavily on the capabilities provided by the [`bd` CLI](../../cli/quickstart.md). Once installed, your agent will be able to:

- Install and instrument the Capture SDK in iOS, Android, and React Native apps
- Add screen tracking, network monitoring, log forwarding, custom fields, and spans to existing integrations
- Authenticate
- Create, deploy, and update workflows
- Query and analyze chart metric data
- Query and analyze session timelines 
- Query issue groups and root-cause crash reports
- Create and manage API and SDK keys
- Query the docs and schema to answer questions

### Examples

**SDK instrumentation**

- *"Install the bitdrift Capture SDK in my Android app and track feature flag exposures."*
- *"Add network monitoring and screen tracking to my iOS app's existing bitdrift integration."*
- *"Set up the bitdrift SDK in my React Native app, including user identity fields after login."*
- *"Validate my existing bitdrift integration and tell me what's missing."*

**Platform observability**

- *"Help me diagnose the top 3 crashes for the latest release of my app over the past week."*
- *"Show me sessions where users experienced sharp memory increases that led to out-of-memory errors for my app."*
- *"Which client API calls are using the most network bandwidth in the latest release of my app?"*
- *"Where are users dropping off the most? Show me sessions with the top user journeys that ended in a force quit for my app."*

!!! info
    When you ask platform observability questions, the agent uses [`bd` CLI](../../cli/quickstart.md) commands to fetch real data from your bitdrift account.

## Why Agent Skills?

By utilizing an on-demand architecture, skills enable extensive functionality while minimizing context window bloat. The agent autonomously retrieves specific skill sets based on task relevance, dynamically injecting the full operational logic and resources into the workflow at the moment of execution.