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.
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-cs-setupgit 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/skills/mims-harvard/tooluniverse/tooluniverse-cs-setup)<a href="https://agentmods.dev/skills/mims-harvard/tooluniverse/tooluniverse-cs-setup"><img src="https://agentmods.dev/badge/skills/mims-harvard/tooluniverse/tooluniverse-cs-setup/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/skills/mims-harvard/tooluniverse/tooluniverse-cs-setup"><img src="https://agentmods.dev/badge/skills/mims-harvard/tooluniverse/tooluniverse-cs-setup.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Rogue Agent · line 3 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
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 | $0.00082 | $0.00933 |
| Opus 5 | $0.00041 | $0.00466 |
| Sonnet 5 | $0.00016 | $0.00187 |
| Haiku 4.5 | $0.00008 | $0.00093 |
Grade A, and why
tooluniverse-cs-setup 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 7d 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 — 63 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Set up ToolUniverse for Claude Science
The upstream ToolUniverse ships a Claude Code plugin (MCP server + uvx + slash commands). Claude Science loads capabilities differently, so this skill installs the equivalent natively: the tooluniverse pip package supplies the 2500+ tools, and the workflow library is packaged into a single dynamically-loaded skill, tooluniverse-research. No uv, no MCP server, no plugin marketplace.
Loading this skill defines tu_build_research_bundle() in the kernel (run cells in the tooluniverse conda env).
Full install / update — four steps
1. Create the conda env (skip if it already exists):
manage_environments(mode="create", name="tooluniverse", python_version="3.11", packages=["pip"])
2. Install (or upgrade) the tools — the pip package is the tool layer:
manage_packages(mode="install", environment="tooluniverse", packages=["tooluniverse"], use_pip=True)
Pin a version for reproducibility with ["tooluniverse==1.3.0"].
3. Stage the workflow bundle — fetch the current repo and rebuild the file tree (run in a python cell, env tooluniverse):
res = tu_build_research_bundle(staging="./tu_staging")
res # {out_dir, n_workflows, n_files, dropped, files_head}
This downloads the repo tarball, parses every tooluniverse-* workflow (dropping the plugin/installer entries), and writes ./tu_staging/out/ = SKILL.md, kernel.py, index.json, workflows/*.md.
4. Publish the skill — push the staged tree into the catalog (run in the repl tool; host.skills.* lives there, not in python):
import os
SKILL = "tooluniverse-research"
out = os.path.abspath("./tu_staging/out")
if any(s["name"] == SKILL for s in host.skills.list()):
host.skills.delete(SKILL) # clean rebuild
for root, _d, fs in os.walk(out):
for f in fs:
p = os.path.join(root, f)
rel = os.path.relpath(p, out)
host.skills.edit(SKILL, rel, open(p, encoding="utf-8").read())
print(host.skills.publish(SKILL, overwrite=True))
(host.skills.publish refuses if kernel.py fails the sidecar gate — the edit result carries the verdict.)
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 7d ago First seen · 63 lines · 82 tokens per session scan A d9a06c9ffd52
tooluniverse-cs-setup is a skill published in the GitHub repository mims-harvard/ToolUniverse (1,680 stars, last pushed yesterday), licensed Apache-2.0. It adds 82 tokens to every session and 933 once invoked, about $0.0004 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 2026-09-03.
Other skills, from other repositories
cost-optimization
Optimize cloud costs across AWS, Azure, GCP, and OCI through resource rightsizing, tagging strategies, reserved instances, and spending analysis. Use when reducing cloud expenses, analyzing infrastructure costs, or implementing cost governance policies.
linkerd-patterns
Implement Linkerd service mesh patterns for lightweight, security-focused service mesh deployments. Use when setting up Linkerd, configuring traffic policies, or implementing zero-trust networking with minimal overhead.
azure-resource-manager-durabletask-dotnet
Azure Resource Manager SDK for Durable Task Scheduler in .NET. Use for MANAGEMENT PLANE operations: creating/managing Durable Task Schedulers, Task Hubs, and retention policies via Azure Resource Manager. Triggers: "Durable Task Scheduler", "create scheduler", "task hub", "DurableTaskSchedulerResource", "provision…
azure-storage-blob-java
Build blob storage applications with Azure Storage Blob SDK for Java. Use when uploading, downloading, or managing files in Azure Blob Storage, working with containers, or implementing streaming data operations.
aws-sdk-java-v2-lambda
Provides AWS Lambda patterns using AWS SDK for Java 2.x. Use when invoking Lambda functions, creating/updating functions, managing function configurations, working with Lambda layers, or integrating Lambda with Spring Boot applications.
aws-cloudformation-dynamodb
Provides AWS CloudFormation patterns for DynamoDB tables, GSIs, LSIs, auto-scaling, and streams. Use when creating DynamoDB tables with CloudFormation, configuring primary keys, local/global secondary indexes, capacity modes (on-demand/provisioned), point-in-time recovery, encryption, TTL, and implementing template…