Skip to content

AI Agent Skill Integration

Library Insight bundles a Custom AI Agent Skill (SKILL.md) that teaches AI assistants (like Claude, Gemini, Cursor, Copilot, Junie, etc.) to verify real dependency APIs before writing code.

The skill is designed around one core rule: do not guess from web examples when the installed library version can be scanned directly. Agents should use search, explain, diff, and ai-export to confirm what exists in the actual artifact.

1. Global Auto-Integration

When you install the CLI globally via the recommended shell installer (install.sh), the script automatically copies the agent skill file into your user profile configurations:

  • ~/.cursor/skills/library-insight
  • ~/.gemini/config/skills/library-insight
  • ~/.claude/skills/library-insight
  • ~/.agents/skills/library-insight
  • ~/.copilot/skills/library-insight
  • ~/.junie/skills/library-insight

Any active AI agent running on your computer will instantly discover and utilize the library-insight command tree when you ask a question.

2. Project Workspace Scoping

If you want to install the skill scoped only to your current project directory (so that any developer working in the repository gets the skill context), run the following command in the project root:

library-insight init
# or
library-insight skills add

This creates .agents/skills/library-insight/SKILL.md inside your project root. AI coding assistants loaded in the workspace will automatically read this skill file to understand how to interact with the Library Insight tool tree.

3. Feeding Source Code Context to LLMs

In addition to compiled dependencies, you can feed raw source code structures directly to AI agents:

  1. Scan your project's local source code: 
    library-insight scan-source app/src/main
  2. Export the index database into a token-optimized directory structure: 
    library-insight ai-export

This generates a compact, token-efficient split structure under build/ai-context/. You can feed this directory into Cursor, Claude, or Gemini workspace chats, giving them full knowledge of your classes, properties, constructors, methods, Javadocs/KDocs, file imports, and declaration line locations without wasting context tokens on massive raw file reads.

4. Real-World Prompt Examples for Developers & AI Agents

Here are concrete examples of how you can prompt AI coding assistants (Cursor, Claude, Gemini, Copilot, etc.) to use Library Insight CLI commands automatically:

Example 1: Exploring Unfamiliar Libraries & DSLs

Developer Prompt: "Use my library-insight CLI to scan androidx.compose.foundation:foundation-layout:1.12.0, search for Grid, explain its DSL structure, and create a working Jetpack Compose Grid layout example."

AI Agent Response Workflow:

  1. Executes library-insight scan androidx.compose.foundation:foundation-layout:1.12.0
  2. Executes library-insight search Grid
  3. Executes library-insight explain Grid --deep to inspect top-level GridKt, receiver scopes (GridScope, GridConfigurationScope), and KDocs in 1 turn.
  4. Generates 100% accurate, hallucination-free Kotlin Compose code matching the exact bytecode API signatures without needing web searches.

Example 2: Scanning Bill of Materials (BOM)

Developer Prompt: "Scan androidx.compose:compose-bom:2024.09.00 using library-insight scan and explain how LazyVerticalGrid and StaggeredGridCells work."

AI Agent Response Workflow:

  1. Executes library-insight scan androidx.compose:compose-bom:2024.09.00 (automatically parses POM XML, filters out non-Android/KMP stubs via default --platform android, and merges target libraries into one index in seconds).
  2. Executes library-insight explain LazyVerticalGrid --deep to discover LazyGridScope, LazyGridItemSpanScope, and GridItemSpan.
  3. Writes precise grid code using the exact version parameter signatures.

Example 3: Auditing Library Version Upgrades & Breaking Changes

Developer Prompt: "Compare Retrofit versions 2.9.0 vs 2.11.0 using library-insight diff and tell me what APIs were added, removed, or deprecated."

AI Agent Response Workflow:

  1. Executes library-insight diff com.squareup.retrofit2:retrofit:2.9.0 com.squareup.retrofit2:retrofit:2.11.0
  2. Analyzes the generated JSON/Console diff report for signature modifications.
  3. Provides a step-by-step migration advisor guide and deprecation checklist.

Example 4: Exporting Local Source Code Context for AI Coding

Developer Prompt: "Scan our local codebase library-insight scan-source app/src/main, run library-insight ai-export, and explain how DashboardViewModel interacts with our repository layer."

AI Agent Response Workflow:

  1. Executes library-insight scan-source app/src/main
  2. Executes library-insight ai-export
  3. Inspects token-optimized markdown files in build/ai-context/ to explain architecture and dependency flows with 0 token waste on raw code files.

Example 5: Version Catalog Auto-Discovery & Zero-Knowledge Scanning

Developer Prompt: "I want to create a grid layout using Compose BOM in our project, but I don't know the exact library version. Find the library name and version from our project's version catalog (gradle/libs.versions.toml), scan it with library-insight, and implement the grid component inside a new file app/src/main/java/com/example/app/ComplexGridLayout.kt."

AI Agent Response Workflow:

  1. Version Catalog Inspection: The AI Agent inspects gradle/libs.versions.toml or build.gradle.kts to locate the project's declared library coordinates or BOM references (e.g. androidx.compose.foundation:foundation-layout or compose-bom = "2024.09.00"). (If not declared in the project, the agent runs library-insight search-central "foundation-layout" to query the latest release on Maven Central).
  2. Auto-Infers Options & Executes Scan: The agent automatically infers missing parameters (such as adding -i "foundation*" or -p android for BOM coordinates) to avoid downloading unnecessary member libraries or platform variants, recommending and executing: 
    library-insight scan androidx.compose:compose-bom:2024.09.00 -i "foundation*" -p android
  3. Mandatory Symbol Search & Explanation: The agent FIRST executes library-insight search Grid to discover all matching symbols across indexed packages, followed by library-insight explain Grid --deep to verify GridScope, GridConfigurationScope, and parameter signatures for that specific version.
  4. Writes Project File: Automatically creates app/src/main/java/com/example/app/ComplexGridLayout.kt with 100% version-matched, working Kotlin code tailored to the developer's project configuration!