CLI and MCP
Scibudy has the same research capabilities through two surfaces:
- CLI commands for shell workflows and scripts.
- A local MCP server for Codex and compatible agents.
The primary commands are scibudy and scibudy-mcp. Compatibility aliases research-cli and research-mcp remain available during v0.x.
CLI command groups
| Goal | Commands |
|---|---|
| Setup and health | setup, doctor, security-audit, bootstrap, install-codex, show-install-state, upgrade-runtime |
| Search | search, source, oa, providers, runs, show |
| Build libraries | collect, download, organize, import-library, libraries, library-show |
| Analyze full text | ingest-library, ingest-item, summarize-library, summarize-item, compare-items, analyze-topic, search-evidence |
| Synthesize | workflow, synthesize-library, profiles, analysis-reports, analysis-report-show |
| Context and UI | bundle-create, bundle-show, ui |
| Journal style | journal-analyze, journal-standardize |
| Local models | install-local-models, warm-local-models, uninstall-local-models, analysis-settings, analysis-update |
Common CLI examples
scibudy doctor --json
scibudy search "simulation-based calibration" --mode general --limit 20
scibudy source openalex "Bayesian atmospheric chemistry" --limit 10
scibudy oa 10.1038/s41467-023-00000-0
scibudy collect "posterior calibration" --target-dir ~/Desktop/posterior-calibration --limit 50
scibudy ingest-library <library_id>
scibudy search-evidence <library_id> "coverage diagnostics"
scibudy synthesize-library <library_id> "simulation-based calibration" --profile auto
scibudy journal-analyze --journal nature-communications --query "atmospheric chemistry Bayesian inference" --target-dir ./nc-style
scibudy journal-standardize --corpus-dir ./nc-style --input ./manuscript.tex
scibudy ui --open
MCP tool families
The research MCP server exposes these main tool families:
| Tool family | Representative tools | Use when |
|---|---|---|
| Search | search_literature, search_biomed, search_source, resolve_open_access |
You need search results or open-access links only. |
| Agent workflow | research_workflow |
You want one call to search, organize, ingest, and synthesize. |
| Library management | collect_library, import_library, list_libraries, read_library, archive_library, restore_library, update_library_item |
You need persistent local libraries. |
| Ingestion and evidence | ingest_library, ingest_library_item, search_library_evidence |
You need to search inside parsed full text. |
| Summaries and synthesis | summarize_library, summarize_library_item, compare_library_items, analyze_library_topic, build_research_synthesis, list_domain_profiles |
You need structured notes or cross-paper synthesis. |
| Reports and bundles | list_analysis_reports, read_analysis_report, read_synthesis_report, generate_context_bundle, read_context_bundle |
You need reusable context for a conversation or manuscript task. |
| Journal style | analyze_journal_style, standardize_journal_text |
You need corpus-based writing-style diagnostics or vocabulary-standardization audits for a target journal. |
| Runtime diagnostics | health_check, security_check, get_analysis_settings, update_analysis_settings |
You need readiness, safety, or backend settings. |
| UI | render_library_manager |
You want the browser or ChatGPT Apps library manager. |
CLI to MCP mapping
| CLI | MCP |
|---|---|
scibudy search |
search_literature |
scibudy collect |
collect_library |
scibudy workflow |
research_workflow |
scibudy ingest-library |
ingest_library |
scibudy search-evidence |
search_library_evidence |
scibudy synthesize-library |
build_research_synthesis |
scibudy journal-analyze |
analyze_journal_style |
scibudy journal-standardize |
standardize_journal_text |
scibudy doctor |
health_check |
scibudy security-audit |
security_check |
Use the CLI when you want reproducible shell commands. Use MCP when Codex should choose tool arguments, inspect results, and continue the research workflow interactively.