ScholarAIO is designed to work best through an AI coding agent, but the CLI remains useful for scripting, inspection, and quick queries.
The authoritative source is always:
scholaraio --help
scholaraio <command> --help
The command groups below are aligned with the current codebase.
scholaraio setup
scholaraio setup check
scholaraio setup agent
scholaraio setup agent --apply
scholaraio setup agent check
setup runs the interactive installation and configuration wizard.setup check reports dependency, parser, API-key, optional service, and runtime directory status.setup agent previews cross-project agent integration for supported coding agents. It covers shell runtime wiring, Codex/OpenClaw skill discovery, project-local wrappers for supported hosts, and Claude Code plugin instructions.setup agent --apply performs the automatic steps. Restart the target agent session afterward so it reloads newly registered skills and wrapper files.setup agent check reports the current agent-integration state without changing files.--target-project writes project-local wrappers with absolute local paths. Review the managed block before committing those files to a shared repository.scholaraio index
scholaraio index --chunks
scholaraio search
scholaraio search --chunk
scholaraio search-author
scholaraio show
scholaraio embed
scholaraio vsearch
scholaraio usearch
scholaraio fsearch
scholaraio top-cited
search performs paper-level keyword search.index --chunks builds a line-addressable evidence chunk index from paper.md and meta.json["toc"].search --chunk searches evidence chunks and returns the source paper, section, line range, and snippet. It supports the normal search filters such as --year, --journal, and --type. This is evidence retrieval, not a knowledge graph.vsearch performs semantic vector search.usearch performs fused keyword + semantic retrieval.fsearch searches across the main library, proceedings, explore databases, and arXiv.show supports layered reading from metadata to full text.scholaraio pipeline [preset]
scholaraio patent-search <query> [--count N]
scholaraio patent-fetch <publication-number-or-url>
scholaraio enrich-toc
scholaraio enrich-l3
scholaraio backfill-abstract
scholaraio refetch
scholaraio translate
scholaraio attach-pdf [--dry-run] [--force]
scholaraio fetch-pdf <doi-or-url-or-title> [--direct] [--out-dir <dir>] [--ingest]
scholaraio fetch-pdf --paper <paper-id> [<paper-id> ...] [--direct] [--force]
scholaraio fetch-pdf --all [--direct] [--force]
pipeline is the main composable ingest entrypoint.fetch-pdf downloads publisher PDFs through the current user network and access context. It does not bypass access controls; use --direct to ignore proxy environment variables such as Clash when the campus network itself has access.fetch-pdf --ingest sends only the fetched PDF into the ingest pipeline. Without --out-dir, the PDF is staged temporarily and is not left in the configured inbox; use --out-dir to keep a separate downloaded copy. If --out-dir is supplied, the PDF is saved there but ingested through an isolated temporary single-file inbox, so unrelated PDFs in that directory are not processed.patent-search discovers patent candidates through USPTO PPUBS by default, with optional ODP API support.patent-fetch downloads a patent PDF into the configured patent inbox for the normal patent ingest flow.refetch refreshes citation counts, bibliographic metadata, and structured references for already ingested papers.refetch --references-only / --refs-only limits the run to DOI papers whose references field is still empty; in single-paper mode it only updates references.attach-pdf attaches a source PDF to an existing paper directory, stores it beside paper.md using the paper directory stem, and regenerates Markdown. It refuses to replace an existing canonical PDF unless --force is supplied.fetch-pdf --paper <id> [<id> ...] re-downloads canonical PDFs for selected existing library papers using source_url or DOI; fetch-pdf --all applies the same logic to the whole library and reports downloaded/skipped/failed counts. Refetching PDFs does not regenerate paper.md; use attach-pdf or the ingest conversion path when Markdown needs to be rebuilt.full, ingest, enrich, and reindex.scholaraio pipeline --help for pipeline options such as --steps, --dry-run, --no-api, and --rebuild.scholaraio refs
scholaraio citing
scholaraio shared-refs
scholaraio topics
scholaraio explore
refs, citing, and shared-refs for citation-graph analysis.topics for BERTopic-based topic modeling and exploration.explore for OpenAlex-backed literature exploration outside the main library.scholaraio import-endnote
scholaraio import-zotero
scholaraio export
scholaraio publish-site
scholaraio gui
scholaraio ws
scholaraio migrate status
scholaraio migrate upgrade --migration-id <id> --confirm
scholaraio migrate verify --migration-id <id>
scholaraio migrate finalize --migration-id <id> --confirm
migrate upgrade is the one-command path from supported legacy layout roots to the current fresh layout. It runs needed store moves, verification, cleanup archival, and final verification in one journal.data/papers/, data/citation_styles/, data/toolref/, data/explore/, data/proceedings/, data/inbox*, data/pending/, workspace/<name>/papers.json, and legacy workspace outputs.migrate finalize remains available when a user or operator has already run store-level migration steps manually and only needs final cleanup and verification.
import-endnote and import-zotero bring existing libraries into ScholarAIO.export handles BibTeX, RIS, Markdown, and DOCX export. BibTeX exports and WebUI Copy BibTeX omit abstracts; stored metadata and RIS abstracts are unchanged.publish-site generates a static site from audited published/*/metadata.json archives, copying PDF/source assets by default and supporting --symlink for local preview.pdf-recovery inspect|export|resolve reviews retained PDF versions and resolves conflicts with snapshot validation. See the Library WebUI guide.gui starts a local, metadata-read-only WebUI with live refresh, audit status, Markdown-rendered abstracts/conclusions, composable metadata filters, server-backed Keyword/Semantic/Unified retrieval for the main library, canonical per-record BibTeX copy, inline PDF preview, and a loopback-only action for opening PDFs in the operating system’s default viewer. On WSL, that action uses a stable Windows edit mirror and automatically reconciles valid PDF saves back to the canonical library copy. The WebUI serves only packaged local assets and does not load remote runtime scripts. See the Library WebUI guide.ws manages paper subsets for focused projects and writing workflows.scholaraio toolref
scholaraio arxiv
scholaraio document
scholaraio diagram
scholaraio style
scholaraio backup
toolref provides versioned scientific tool documentation lookup.toolref subcommands are fetch, show, search, list, and use.arxiv supports arXiv search and PDF fetch.document provides Office-document utilities such as inspection.diagram generates editable scientific diagrams from paper content or structured text. See the Graphviz Diagram Guide for DOT/SVG workflows.style manages citation styles.backup lists configured rsync targets, runs named backup plans, and restores full instance backups.scope: data retain the legacy configured-directory sync. Targets with scope: instance preserve config files, data, workspaces, published archives, and instance control metadata under their normal relative paths.backup.backup restore <target> --destination <dir> validates the remote instance manifest before restoring. Non-empty destinations require --force.BatchMode=yes; password targets use the internal askpass path instead of waiting for a terminal prompt.config.local.yaml when present. Secrets are encrypted in transit by SSH but remain plaintext files on remote storage; environment-only API keys are not captured.ssh-keyscan ... >> ~/.ssh/known_hosts, then ssh -i <key> -p <port> <user>@<host> true, then scholaraio backup run <target> --dry-run.scholaraio audit
scholaraio repair
scholaraio rename
scholaraio setup
scholaraio insights
scholaraio metrics
scholaraio proceedings
scholaraio citation-check
audit checks missing metadata, duplicate DOIs, filename issues, and title/content mismatches.audit uses paper-type-aware skips so documents, patents, dissertations, and similar front matter do not create spurious title_mismatch warnings.audit reports noncanonical_paper_type when a stored type differs from the value used by search and library views. Metadata writes normalize aliases such as jour to journal-article and monograph to book. Missing types with both a journal and DOI default to journal-article; selected review venues infer review for missing or generic article types. Venue inference is a heuristic, and explicit specialized types such as editorial or correction are preserved. Existing records are not rewritten by an audit or search.setup is the environment check and setup wizard entrypoint.insights analyzes research behavior such as hot keywords and reading trends.metrics shows LLM token and runtime usage.proceedings provides dedicated proceedings helpers.citation-check verifies whether citations in text are backed by the local library.Use the agent for the full workflow, and fall back to CLI commands when you want: