wiki_ingestlisted

# LLM Wiki — Ingest Skill (wiki_ingest) > **CLI (read first):** This skill drives the `magi` CLI (MAGI research workspace tool, assumed installed on PATH). If unsure of your surroundings, run `magi sync` first to locate the workspace. For the full syntax of any command: `magi <command> --help`. This skill handles converting external material (URLs, PDFs, local text files, and items inside `inbox/`) into raw sources. > **Tooling (framework-agnostic):** This skill is written tool-agnostic. Map each capability to your own agent's tool — *read-file* (`Read` in Claude Code, `view_file` in Antigravity), *sub-agent / parallel task* (`Task`/`Agent` in Claude Code, `invoke_subagent` in Antigravity), *shell* (`Bash`/`PowerShell`). Use the closest equivalent your framework provides; if a parallel sub-agent tool is unavailable, transcribe PDF pages sequentially yourself (still verifying the full page count). When the user asks to ingest documents (or runs the command without a path): 1. **Resolve Ingestion Targets**: * If a specific file path or URL is provided, process that target. * If NO target is provided, automatically scan the `inbox/` directory. * If `inbox/` contains multiple files, you **MUST** loop through all of them and process them one by one in a batch. * If `inbox/` is empty, only then prompt the user to specify a file or source URL. 2. **Identify Source Type**: For each target file, academic papers go to `raw/papers/`, web pages to `raw/artic