--- title: "AI-driven instrumentation made possible with new mobile SDK skill" metaTitle: "AI-driven instrumentation with new mobile SDK skill" slug: "skill-that-instruments-mobile-sdk" blurb: "Introducing bd-instrumentation: a new agent skill that handles installation and instrumentation of the bitdrift Capture SDK." metaDescription: "New agent skill handles installation and instrumentation of the bitdrift Capture SDK. bd-instrumentation, bd-cli and bd-docs empower agentic mobile observability." cover: url: "/assets/posts/skill-that-instruments-mobile-sdk/feature-hero-desktop@1x.webp" alt: "Illustration representing AI-driven instrumentation for mobile SDK integrations." socialThumbnail: url: "/assets/posts/skill-that-instruments-mobile-sdk/feature-hero-desktop@1x.webp" alt: "Illustration representing AI-driven instrumentation for mobile SDK integrations." author: - "collin" tags: - "observability" - "mobile" publishedDate: "2026-05-28T00:00:00Z" modifiedDate: "2026-05-28T00:00:00Z" --- Introducing bd-instrumentation: a new agent skill that handles installation and instrumentation of the bitdrift Capture SDK. We recently released two things in quick succession: 1. The [**bitdrift Public API**](https://blog.bitdrift.io/post/public-api): The foundation that makes mobile observability programmable and agent-ready via the bitdrift CLI. 2. Then, [**bd skills**](https://blog.bitdrift.io/post/query-reality-ai-observability): a set of agent skills that give AI coding agents access to everything happening on your customers' devices in real time, so they can triage, investigate, and debug mobile issues using bitdrift’s mobile observability platform, without a human in the loop. Not surprisingly, the [investigation skills (bd-cli and bd-docs)](https://docs.bitdrift.io/product/skills/overview) gained traction quickly. Developers want agents that can query production users (or [reality](https://blog.bitdrift.io/post/query-reality-ai-observability), as we like to call it), tail device logs, and answer questions about production behavior. It's agentic observability at its finest. But today we’re shining a light on [bd-instrumentation](https://docs.bitdrift.io/product/skills/overview#bd-instrumentation): the agent skill that handles install and instrumentation of the bitdrift Capture SDK. ## Because you can't investigate what you haven't instrumented Before an agent can query a session, triage a crash, or tail logs from a specific device, the bitdrift [Capture SDK](https://bitdrift.io/feature/performance-centric) has to be in your app. Getting it there properly and completely has always been the unglamorous prerequisite that everyone wishes would just handle itself. It's not a hard problem in the abstract. [The docs are clear.](https://docs.bitdrift.io/sdk/quickstart) The APIs are well designed. But in practice, mobile app instrumentation requires careful platform-specific judgment: the right dependency management approach for your build system, the right networking integration for your HTTP client, the right setup for [crash reporting](https://bitdrift.io/use-cases/crash-reporting), screen view tracking, and session attribution. Get it half right and you get half the value. And "half right" after spending three weeks in the backlog is where many teams land with observability SDKs in general. bd-instrumentation brings a sigh of relief to the process. It encodes what a complete, correct integration looks like, and hands it to the agent. For both iOS and Android, it covers: * **Adding the Capture SDK** dependency (Gradle, Swift Package Manager, or CocoaPods) * **Initializing the logger** with your project's SDK key * **Wiring up networking** integrations (OkHttp on Android, URLSession on iOS) * **Instrumenting** screen views and navigation events * **Platform-specific extras:** Gradle plugin for automatic WebView instrumentation, ProGuard mapping uploads for deobfuscated crash reports The agent reads your codebase, understands what frameworks you're already using, and instruments accordingly. It doesn't stop at the dependency. It doesn't forget the network interceptor. It doesn't leave you with a half-wired SDK that technically compiles but doesn't tell you anything useful. ## The bigger picture In the old world, mobile instrumentation required precious engineering cycles, got done inconsistently, and blocked observability value for weeks. Slow app release cycles and challenging SDK syncs limited the telemetry being collected. But observability no longer has to be a reactive exercise. With agents like Cursor and Claude Code, "we don’t have the engineering bandwidth" stops being a valid excuse. Rather than drowning in a sea of alerts, charts, and logs, agents can autonomously triage, investigate, debug, and even fix issues as they happen. The real unlock is whether you can teach the agent what the best solutions engineer at your company knows; one that works on every PR, never goes on vacation, and never forgets the naming convention. Agent skills make this possible, and the bitdrift instrumentation skill is the entry point. ## Closing the loop Together, bd-instrumentation, bd-cli, and bd-docs bring agentic mobile observability to life: 1. **Install and instrument** the Capture SDK: your agent handles it, correctly, from day one 2. **Investigate:** Your agent queries live sessions, tails device logs, surfaces what's happening on real devices right now 3. **Iterate:** Because the bitdrift platform is controlled dynamically, agents can adjust what telemetry gets collected without waiting for an app release Agents can dynamically create and collect telemetry on any device instantly. User journeys, performance metrics, and behavioral changes are all available when the agent needs them, not ten days later... waiting for an app release. ## What are you waiting for? Get started with bd skills and give it a try. Install all three skills with one command: ```shell npx skills add bitdriftlabs/bd-skills ``` Compatible with Claude Code, Cursor, Codex, Copilot, and any agent that supports the open [agentskills.io](https://agentskills.io/) standard. Full documentation at [docs.bitdrift.io/product/skills/overview](https://docs.bitdrift.io/product/skills/overview). New to bitdrift’s mobile observability platform? [Check out the sandbox](https://bitdrift.io/sandbox) to see it for yourself, or [get a live demo](https://bitdrift.io/contact-us) and we’ll show you around. --- ## Frequently asked questions ### What is agentic observability? Agentic observability is the practice of using AI agents to autonomously monitor, investigate, and debug issues in your application, rather than relying on engineers to manually sift through dashboards, alerts, and logs. Instead of observability being a reactive exercise (something goes wrong, someone digs in), agents can triage and investigate in real time, surfacing answers without a human in the loop. For mobile engineering teams specifically, this requires access to high-fidelity, unsampled device data, which bitdrift’s mobile observability platform uniquely provides. ### What is an observability SDK? An observability SDK is a package of code you integrate into your app that collects and reports data about how your app is behaving in production: things like crashes, network requests, user sessions, performance metrics, and log output. Rather than finding out about problems from user complaints or app store reviews, an observability SDK gives you visibility into what's actually happening on real devices, in real time. The bitdrift Capture SDK is built specifically for mobile, with the ability to dynamically adjust what gets collected without shipping a new app release. ### What is bd-instrumentation? bd-instrumentation is an agent skill that teaches your AI coding agent how to install and instrument the bitdrift Capture SDK in your iOS or Android app. It handles everything from adding the dependency to wiring up networking integrations, crash reporting, and screen view tracking — correctly, for your specific codebase. ### Which AI coding agents does bd-instrumentation work with? It works with any agent that supports the open agentskills.io standard, including Claude Code, Cursor, Codex, and Copilot. ### How is bd-instrumentation different from the other bd skills? bd-cli and bd-docs are for investigation: querying live sessions, tailing device logs, and answering questions about production behavior. bd-instrumentation handles the prerequisite: getting the Capture SDK properly installed and instrumented so there's something to investigate in the first place. ### Do I need all three skills, or can I install bd-instrumentation on its own? You can install individual skills, but running `npx skills add bitdriftlabs/bd-skills` installs all three at once. If you're starting a new integration, bd-instrumentation is the natural first step. bd-cli and bd-docs are only useful once your app is instrumented.