tooluniverse-cs-setup

Install or update ToolUniverse in Claude Science — create the conda env, install the tooluniverse pip package, and (re)build the tooluniverse-research skill by fetching the current workflow library from GitHub. Use for first-time setup, upgrading the ToolUniverse version, refreshing the bundled workflows after an upstream release, or reinstalling on a new machine.

Quality

94%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

86%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A tight, highly actionable setup guide: all four steps are executable as written, sequencing is clear with per-step environment guidance, and the Notes section captures genuinely non-obvious environment specifics (read-only cache, what is not ported). The only gaps are minor — no explicit error-recovery loop around the destructive republish step, and a couple of sentences that could be trimmed.

Suggestions

Add a brief error-recovery note to step 4: what to do when host.skills.publish fails the kernel.py sidecar gate (e.g., read the verdict from the edit result, fix kernel.py in ./tu_staging/out, re-run the publish loop).

Trim the "Updating" note to a single sentence pointing at steps 2–4, and move the tooluniverse==1.3.0 version pin out of the main flow (or mark it as an example) so the pinned version doesn't read as current guidance.

Dimension	Reasoning	Score
Conciseness	The body is dense and environment-specific (Claude Science loading model, sandbox cache, sidecar gate) with no re-teaching of concepts Claude already knows. Not 5: the "Updating: rerun steps 2–4" note partially restates the step list, and the version-pin example "tooluniverse==1.3.0" is mildly time-sensitive without a deprecation context.	4 / 5
Actionability	Every step is copy-paste ready: complete manage_environments/manage_packages calls, the tu_build_research_bundle invocation with its expected output dict, a full publish script, and a Verify section with a concrete tool call and expected result. Fully executable with no gaps.	5 / 5
Workflow Clarity	Four clearly sequenced steps with skip conditions ("skip if it already exists"), per-step environment labels, and a dedicated Verify section with expected output. Not 5: step 4 performs a destructive clean rebuild (delete + republish) and mentions the sidecar-gate refusal, but there is no explicit error-recovery loop (fix and retry) if publishing or verification fails.	4 / 5
Progressive Disclosure	No bundle files exist (references/, scripts/, assets/ are absent) and none are needed: the ~57-line body is a single self-contained set of setup instructions with well-organized sections (steps, Verify, Notes), no nested references, and no content that belongs in a separate file. Well-organized sections suffice here.	5 / 5
	Total	18 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

An exemplary description: concrete multi-action capability statement in third person, followed by an explicit "Use for" clause enumerating four distinct trigger scenarios with natural synonyms. Both what and when are unambiguous, and the tool/platform naming eliminates conflict risk.

Dimension	Reasoning	Score
Specificity	The description lists multiple concrete actions — "create the conda env, install the tooluniverse pip package, and (re)build the tooluniverse-research skill by fetching the current workflow library from GitHub" — comprehensively covering everything the skill does. It exceeds anchor 4 because there are no minor gaps in the action coverage for this task.	5 / 5
Completeness	It explicitly answers both questions: what ("Install or update ToolUniverse… create the conda env, install the tooluniverse pip package, and (re)build the tooluniverse-research skill") and when ("Use for first-time setup, upgrading…, refreshing…, or reinstalling on a new machine") with concrete trigger phrases. This matches the anchor 5 example structure exactly; score 4 would require a less explicit when-clause.	5 / 5
Trigger Term Quality	Trigger phrases "Use for first-time setup, upgrading the ToolUniverse version, refreshing the bundled workflows after an upstream release, or reinstalling on a new machine" enumerate the natural ways a user would express the need, with synonym coverage (install/update, setup, upgrade, refresh, reinstall). Anchor 4 ("a few natural terms missing") does not fit — the usage space is fully covered.	5 / 5
Distinctiveness Conflict Risk	"ToolUniverse in Claude Science" names a specific tool and platform — a clear niche with distinct triggers that would not plausibly fire for any other skill. Conflict risk is minimal.	5 / 5
	Total	20 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: mims-harvard/ToolUniverse
Path: skills/tooluniverse-cs-setup/SKILL.md
Commit: 14c4186

Reviewed: 8 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.