ToolUniverse is a collection of tools, interfaces, and supporting components for building AI systems that perform scientific work. It is for developers creating AI scientist agents that use APIs, databases, machine-learning tools, and domain-specific utilities. The catalogue includes skills, commands, an MCP server, an agent, and a hook for working with the ecosystem.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/mims-harvard/ToolUniverseWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/mims-harvard/tooluniverse/literature-sweep)<a href="https://agentmods.dev/commands/mims-harvard/tooluniverse/literature-sweep"><img src="https://agentmods.dev/badge/commands/mims-harvard/tooluniverse/literature-sweep/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/mims-harvard/tooluniverse/literature-sweep"><img src="https://agentmods.dev/badge/commands/mims-harvard/tooluniverse/literature-sweep.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00144 | $0.02494 |
| Opus 5 | $0.00072 | $0.01247 |
| Sonnet 5 | $0.00029 | $0.00499 |
| Haiku 4.5 | $0.00014 | $0.00249 |
Grade A, and why
literature-sweep scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 179 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run a literature sweep on this topic: $ARGUMENTS
A raw PubMed_search dump is overwhelming and often noisy (many hits don't
actually address the topic). Curate down to a small, high-quality reading
list with structured metadata.
Process
1. Plan the query strategy
Decompose the topic into 1-3 search formulations. Different sources accept different syntax, so you'll often run the same idea multiple ways.
For "KRAS G12C inhibitor resistance mechanisms":
- Strict:
(KRAS[ti] OR sotorasib OR adagrasib) AND resistance - Broad:
KRAS G12C resistance mechanism - Mechanism-focused:
KRAS G12C bypass OR rebound
State the formulations in one line before searching.
2. Search several independent literature sources (adaptive by domain)
ToolUniverse exposes 15+ keyword-searchable literature indexes. Each maintains its own coverage, so running a topic across several catches papers any single one misses. Don't blindly fire all of them — that's slow and noisy. Pick the always-on CORE set, then add the domain rows that match the topic.
ALWAYS run (multi-field core — 4 indexes):
# PubMed (NIH; peer-reviewed biomedical, MeSH-indexed)
tu run PubMed_search_articles '{"query":"KRAS G12C resistance mechanism","limit":20}'
# EuropePMC (broader: clinical, agricultural, pharma + preprints via SRC:PPR)
tu run EuropePMC_search_articles '{"query":"KRAS G12C resistance mechanism","limit":20}'
# OpenAlex (250M+ works, every discipline; good cross-field recall)
tu run openalex_search_works '{"search":"KRAS G12C inhibitor resistance","per_page":20}'
# Semantic Scholar (AI-ranked citation graph; catches non-MeSH-indexed work)
tu run SemanticScholar_search_papers '{"query":"KRAS G12C inhibitor resistance","limit":20}'
THEN add domain-specific indexes when the topic matches:
| Topic signal | Add these sources | Why |
|---|---|---|
| Biomedical / clinical / gene·drug·disease | PMC_search_papers (full text), PubTator3_LiteratureSearch (entity & relation queries, e.g. relations:treat|@CHEMICAL_X|@DISEASE_Y) |
Full-text body hits + entity-normalized recall |
| Clinical practice / treatment guidelines | PubMed_Guidelines_Search |
Filters to guideline / practice-guideline pub types |
| CS / ML / AI / algorithms | ArXiv_search_papers, DBLP_search_publications |
arXiv preprints + CS bibliography (often not in PubMed) |
| Physics / HEP / astro | InspireHEP_search_papers |
1.6M+ particle/astro physics records |
| Broad / cross-disciplinary / hard-to-find | Crossref_search_works, CORE_search_papers, DOAJ_search_articles, Fatcat_search_scholar |
DOI registry + open-access aggregators + Internet Archive Scholar |
| Need the very latest (preprints) | EuropePMC_search_articles with SRC:PPR, OSF_search_preprints |
bioRxiv/medRxiv/PsyArXiv etc. before peer review |
| Datasets / code / supplementary outputs | Figshare_search_articles, Zenodo_search_records |
Research data and software with citable DOIs |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 179 lines · 144 tokens per session scan A cdcde6451cba
literature-sweep is a command published in the GitHub repository mims-harvard/ToolUniverse (1,680 stars, last pushed yesterday), licensed Apache-2.0. It adds 144 tokens to every session and 2,494 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
template
Manage issue templates for streamlined issue creation.
discover
Search the Everruns(Dev) API catalog for available operations.
doctor
Badi configuration validation. Checks all Badi components and produces a diagnostic report.
swarm-status
Command documentation for swarm-status in category swarm.
archive-ledger
../../../shared/commands/archive-ledger.md.
research
Document codebase as-is with thoughts directory for historical context.