CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

tooluniverse-codex-plugin

Install, set up, verify, update, pin, uninstall, or troubleshoot the ToolUniverse plugin on OpenAI Codex. ALWAYS consult this skill for any of those — don't answer from memory, because the exact marketplace name (mims-harvard/ToolUniverse), the "codex plugin marketplace add" then "codex plugin add -m tooluniverse" flow, Codex's startup auto-upgrade behavior, the uvx tooluniverse MCP server, and the API-key env vars are easy to get wrong. Use it whenever someone wants to get ToolUniverse (or "the 2,700+ scientific tools" / "the harvard tools") working on Codex, says the Codex plugin or its tools/skills won't load, hits a uvx or MCP-server startup error, asks how Codex updates it, wants to pin or remove it, or finds it running an old tool version — even if they never say the word "plugin". Not for the Claude Code plugin (use tooluniverse-claude-code-plugin), for running research with the tools, or for authoring new tools or skills.

Quality

94%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

88%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 highly actionable, well-sequenced install guide: every operation is a copy-paste command, verification steps state expected outputs, and the diagnosis section provides a genuine check→fix→restart feedback loop. It is slightly over-inclusive for its stated user-facing purpose — the maintainer section and thrice-repeated cache-clean advice are the main trimmable tokens, and the one external reference points into another skill's bundle.

Suggestions

Move the "For plugin maintainers" section into a separate reference file (e.g. MAINTAINERS.md) — it targets a different audience than the user-facing install flow and adds ~10 lines of off-purpose context to the main skill body.

Deduplicate the `uv cache clean tooluniverse` advice, which appears in Updates, Troubleshooting, and step 7 of the diagnosis script — keep one canonical mention and cross-reference it from the other two spots.

Make the API_KEYS_REFERENCE.md pointer resolvable within this skill's own bundle (or inline the few most common keys like NCBI_API_KEY), since it currently depends on the separate setup-tooluniverse skill being installed.

DimensionReasoningScore

Conciseness

The body is command-first and tight — every step is a copy-paste block or a dense table, with no explanation of concepts Claude already knows. It falls short of 5 because the "For plugin maintainers" section (~10 lines about sync-skills.sh and release scripts) serves a different audience than the stated user-facing install flow, and `uv cache clean tooluniverse` is repeated in three places (Updates, Troubleshooting, and step 7 of the diagnosis script). Not 3: the redundancy is deliberate triage structure, not padded prose, and the bulk of the document is lean.

4 / 5

Actionability

Fully executable throughout: "codex plugin marketplace add mims-harvard/ToolUniverse", "codex plugin add tooluniverse -m tooluniverse", "codex plugin list --json", the export block with "...then start codex from that same shell", the pin snippet with the exact "args": ["tooluniverse@1.3.0"] line and where to find the file, and a 7-step diagnosis script with inline FIX commands. Not 4: even edge cases (shadowing global install, Python too old, air-gapped install) get exact commands rather than hints.

5 / 5

Workflow Clarity

Clear sequence — Prerequisites ("check once") → Install (two numbered commands) → restart → "Verify it worked" with expected output ("expect: tooluniverse (enabled)") — plus an explicit error-recovery feedback loop in "Automated diagnosis & repair": "apply the `FIX` for whatever fails, then restart Codex". Not 4: validation checkpoints state expected results, and recovery is a numbered check→fix→retry checklist rather than an implicit step.

5 / 5

Progressive Disclosure

No bundle files exist for this skill (no references/, scripts/, or assets/), so this scores structure alone: a single ~163-line file with well-labeled sections, and the one deferred item is clearly signaled — "Full key list: the bundled `setup-tooluniverse` skill → `API_KEYS_REFERENCE.md`" — keeping a long key list out of the main file. Not 5: that pointer targets another skill's bundle rather than a file in this skill, the maintainer section is inlined content that belongs in a separate reference, and the single file carries more than a lean overview needs. Not 3: navigation is easy, sections are well-organized, and the one reference is one level deep and explicit.

4 / 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: it names the full lifecycle of concrete actions, includes the natural phrases and nicknames users would actually say (down to handling those who never say "plugin"), gives an explicit use-when clause with concrete triggers, and closes with explicit negative boundaries that separate it from sibling skills. Voice is appropriately third-person/imperative throughout.

DimensionReasoningScore

Specificity

The first sentence enumerates the full plugin lifecycle — "Install, set up, verify, update, pin, uninstall, or troubleshoot the ToolUniverse plugin on OpenAI Codex" — concrete actions with comprehensive coverage from install through uninstall. Not 4: there are no coverage gaps; every lifecycle operation is explicitly named.

5 / 5

Completeness

Explicitly answers both parts: the 'what' is the concrete action list in the first sentence, and the 'when' is "Use it whenever someone wants to get ToolUniverse... working on Codex, says the Codex plugin or its tools/skills won't load, hits a uvx or MCP-server startup error, asks how Codex updates it, wants to pin or remove it, or finds it running an old tool version" — concrete trigger phrases, plus negative boundaries. Not 4: the 'when' is explicit and enumerates specific triggering scenarios rather than being generic.

5 / 5

Trigger Term Quality

Covers natural user phrasings comprehensively, including synonyms and colloquial aliases: "get ToolUniverse (or \"the 2,700+ scientific tools\" / \"the harvard tools\") working on Codex", "says the Codex plugin or its tools/skills won't load", "hits a uvx or MCP-server startup error", "finds it running an old tool version", and explicitly handles users who "never say the word 'plugin'". Not 4: it includes the paraphrases and nicknames users would actually say, not just good keywords.

5 / 5

Distinctiveness Conflict Risk

Clear niche (the Codex plugin lifecycle only) with explicit exclusions: "Not for the Claude Code plugin (use tooluniverse-claude-code-plugin), for running research with the tools, or for authoring new tools or skills". Not 4: it actively disambiguates against the most likely conflicting sibling skill, leaving minimal overlap risk.

5 / 5

Total

20

/

20

Passed

Validation

93%

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

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

referenced_paths_exist

Referenced path issues: 1 missing

Warning

Total

15

/

16

Passed

Repository

mims-harvard/ToolUniverse

Path

skills/tooluniverse-codex-plugin/SKILL.md

Reviewed

Table of Contents

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.