Skip to content

AI Context & Integration

Splits the scanned database into a token-efficient directory structure under build/ai-context/. AI agents read metadata.json first, then load only the class files they need — reducing token usage by 95%+.

library-insight ai-export

Positional Arguments:

  • [output-dir]: Optional target output directory to save AI context files (default: build/ai-context/)

Optional Parameters:

  • --db <file>: Index database JSON file path to read from (default: build/library-insight-index.json)

Example with options:

library-insight ai-export custom-ai-context/ --db custom-index.json

Example output:

Generated compact LLM context directory structure at: build/ai-context

mcp — MCP Server

Start the Model Context Protocol server on stdio. Connect Cursor, Claude Desktop, or any MCP-compatible IDE to use scan_library, search_symbols, explain_class, and dsl_report tools natively.

library-insight mcp

Optional Parameters:

  • --db <file>: Index database JSON file path to read from and write to (default: build/library-insight-index.json)

Example with options:

library-insight mcp --db /path/to/project/custom-index.json

Example output (JSON-RPC tools list response):

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "tools": [
      {
        "name": "scan_library",
        "description": "Scans a Java/Kotlin library and creates an API index.",
        "inputSchema": {
          "type": "object",
          "properties": { "pathOrCoordinate": { "type": "string" } }
        }
      },
      {
        "name": "search_symbols",
        "description": "Search for symbols in the active library index.",
        "inputSchema": {
          "type": "object",
          "properties": { "query": { "type": "string" } }
        }
      }
    ]
  }
}

MCP vs CLI: If an MCP server is already configured in your IDE, prefer it over running CLI commands directly.

init — Initialize Workspace Skill

Write a SKILL.md into .agents/skills/library-insight/ so local AI agents auto-discover the CLI.

library-insight init

Example output:

Initializing Library Insight agent environment...
Creating directory: .agents/skills/library-insight/
Successfully initialized workspace skill instructions!

skills — Manage Agent Skills

Add skill to the current workspace:

library-insight skills add

List registered skills in the current workspace:

library-insight skills list

Example output (skills list):

Workspace AI Agent Skills:
  - [Installed] library-insight

export — Export Index

Export the scanned index to Markdown or JSON.

For large libraries, Markdown files can be huge. Use ai-export for AI prompts instead.

Export to Markdown format:

library-insight export markdown

Export to JSON format:

library-insight export json

Positional Arguments:

  • [output-file]: Optional target output file path to write export content to. If not specified, defaults to build/API_REFERENCE.md or build/library-insight-index.json. Use - to print to stdout.

Optional Parameters:

  • --db <file>: Index database JSON file path to read from (default: build/library-insight-index.json)

Example with options:

library-insight export markdown API_REFERENCE.md --db custom-index.json

Example output:

Exported MARKDOWN to: build/API_REFERENCE.md