connector-initlisted

# Role Socratic interview → generates `<run-path>/connector.md` tailored to the user's source. Use when templates in `${CLAUDE_PLUGIN_ROOT}/templates/connectors/` don't match. # Invocation /ultra-analyzer:connector-init <run-path> # Protocol ## Step 1: Pre-flight - Verify `<run-path>/state.json` exists (run is initialized). - If `<run-path>/connector.md` already exists, ask: overwrite, or keep existing? ## Step 2: Interview (use AskUserQuestion) Ask these questions one at a time, in order. Each answer shapes subsequent questions. **Q1: Source category** - What kind of data source are you analyzing? Options: database (Mongo/Postgres/SQLite/etc.) | filesystem (files on disk) | HTTP API (REST/GraphQL) | browser (scraping/automation) | streaming (logs, Kafka) | archive (tarball, zip of structured files) | custom/other **Q2: Access mechanism** - How does the pipeline connect to this source? - If database: connection string (env var name?) or MCP tool name? - If API: base URL, auth mechanism (bearer token? OAuth? API key header?) - If filesystem: root path, file globs? - If browser: playwright-persistent MCP, or browsermcp MCP, or CLI tool? - If custom: describe the CLI / library / process that reads it **Q3: Unit of analysis** - What is a single "unit" in this source? - Database: a collection/table - Filesystem: a file (or a directory) - API: an endpoint or a resource collection - Browser: a page or a DOM subtree - Streaming: a log file or a time wi