CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

setup-tooluniverse

Install and configure ToolUniverse for any use case — MCP server (chat-based), CLI (command line with 14 subcommands), or Python SDK (Coding API with 3 calling patterns). Covers uv/uvx setup, MCP configuration for 12+ AI clients (Cursor, Claude Desktop, Windsurf, VS Code, Codex, Gemini CLI, Trae, Cline, etc.), full CLI reference (tu list/grep/info/find/run/test/status/build/remote/doctor/serve/connect/connections/disconnect), Coding API quickstart, agentic tools, code executor, API key walkthrough, skill installation, and upgrading. Use when user asks how to set up ToolUniverse, which access mode to use (MCP vs CLI vs SDK), configuring MCP servers, using the CLI, troubleshooting installation, upgrading, or mentions installing ToolUniverse or setting up scientific tools. Also triggers for "how do I use ToolUniverse", "what's the best way to access tools", "command line", "tu command", "coding API", "tu build".

Quality

90%

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

The canonical home for this skill is setup-tooluniverse in mims-harvard/ToolUniverse

SKILL.md
Quality
Evals
Security

Quality

Content

77%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.

The body is a highly actionable, well-sequenced setup guide with executable commands, validation checkpoints, and error-recovery guidance for all three access modes. Its weaknesses are moderate over-explanation of concepts Claude already knows (reassurance about the curl command, the config-file explainer), a dead reference to API_KEYS_REFERENCE.md, and four bundle scripts that are never surfaced, leaving progressive disclosure underdeveloped.

Suggestions

Trim user-reassurance asides Claude doesn't need — e.g. '(This is a safe, standard command that downloads and installs uv...)' and the 'Config files are plain text that store settings' explainer — and let the agent improvise user-facing framing.

Fix the dead reference: either add API_KEYS_REFERENCE.md to the bundle or point the link at an existing file/path (e.g. a references/ file), since an agent following it now hits a missing file.

Surface the scripts/ bundle in the body (e.g. 'Run `python scripts/verify_installation.py` to confirm the install' or 'Use `python scripts/diagnose_setup.py` when setup fails') so the four existing diagnostic scripts are discoverable.

DimensionReasoningScore

Conciseness

Quotes: "(This is a safe, standard command that downloads and installs `uv`, a small package manager. It's widely used by Python developers...)" and "Config files are plain text that store settings — like a preference list for the app. You don't need to understand the format" — both explain concepts Claude already knows, and the "universal remote for scientific databases" analogy adds padding. Mostly efficient (the tables and command blocks are dense and useful) but includes unnecessary explanation that could be trimmed, matching anchor 3; not 2 because the majority of the 440 lines are concrete commands, paths, and tables rather than padded prose.

3 / 5

Actionability

Quotes: executable install command "curl -LsSf https://astral.sh/uv/install.sh | sh", ready-to-run CLI examples like "uvx --from tooluniverse tu run FAERS_count_death_related_by_drug '{\"medicinalproduct\": \"metformin\"}'", complete SDK snippets ("from tooluniverse.tools import UniProt_get_entry_by_accession"), and a paste-ready JSON config block with a validation command ("python3 -m json.tool < <config>"). Copy-paste ready commands cover all three access modes and common cases — the level-5 anchor; level 4 would imply missing key details, and none are missing.

5 / 5

Workflow Clarity

Quotes: explicit checkpoints "Verify: `uv --version`", "Then validate before restarting the app: python3 -m json.tool...", and a dedicated test phase "Step 4: Test Together — Don't just tell — do it WITH the user" with concrete test calls ("execute_tool(\"PubMed_search_articles\", ...)"). The sequence (Step 1 choose mode → Step 2 uv → mode-specific setup → Step 3 keys → Step 4 test → Step 5 skills) is clear, with a feedback loop via the Common Issues table ("If issues: Most common: app not restarted, `uv` not in PATH...") and safety checks on semi-destructive steps ("Append (don't overwrite). Check for existing section first.") — matches the level-5 anchor with validation, recovery, and checklists.

5 / 5

Progressive Disclosure

The one well-signaled reference — "See [references/mcp-configs.md](references/mcp-configs.md) for these" — exists and is one level deep, but "See [API_KEYS_REFERENCE.md](API_KEYS_REFERENCE.md) for the complete list" points to a file that does not exist anywhere in the bundle (a dead reference that breaks navigation), and the four scripts in scripts/ (check_prerequisites.py, diagnose_setup.py, list_tool_categories.py, verify_installation.py) are never mentioned in the body, so bundle value is undiscoverable. Structure exists but with a broken pointer and orphaned bundle files — anchor 3 ('references present but not clearly signaled; content that should be separate is inline') fits better than anchor 4's 'minor organization gaps', since a nonexistent referenced file is a real navigation failure, not a minor gap.

3 / 5

Total

16

/

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.

The description is exemplary across all dimensions: it enumerates concrete capabilities in third person, provides explicit 'Use when' guidance with verbatim trigger phrases and synonyms, and is tightly scoped to a distinct product niche. Its only weakness is length (~950 characters of dense clauses), which borders on verbosity, though every clause carries specific information rather than filler.

DimensionReasoningScore

Specificity

Quotes: "Install and configure ToolUniverse for any use case — MCP server (chat-based), CLI (command line with 14 subcommands), or Python SDK (Coding API with 3 calling patterns)" and "full CLI reference (tu list/grep/info/find/run/test/status/build/remote/doctor/serve/connect/connections/disconnect)". Multiple specific concrete actions with comprehensive coverage (uv/uvx setup, MCP config for 12+ named clients, API key walkthrough, upgrading), matching the level-5 anchor rather than level 4's 'minor gaps in coverage'. Voice is third person ('Install and configure'), so no voice penalty applies.

5 / 5

Completeness

Quotes: explicit what ("Covers uv/uvx setup, MCP configuration for 12+ AI clients... full CLI reference... Coding API quickstart, agentic tools, code executor, API key walkthrough, skill installation, and upgrading") followed by explicit when with concrete trigger phrases ("Use when user asks how to set up ToolUniverse..."). Both 'what' and 'when' are clearly and explicitly answered with concrete triggers — the level-5 example pattern exactly; level 4 would require the 'when' to be less explicit.

5 / 5

Trigger Term Quality

Quotes: "Use when user asks how to set up ToolUniverse, which access mode to use (MCP vs CLI vs SDK), configuring MCP servers, using the CLI, troubleshooting installation, upgrading" and "Also triggers for \"how do I use ToolUniverse\", \"what's the best way to access tools\", \"command line\", \"tu command\", \"coding API\", \"tu build\"". Comprehensive natural-term coverage including verbatim user phrasings, synonyms ('command line', 'tu command'), and specific command names, matching the level-5 anchor; not level 4 since no common natural variations are missing.

5 / 5

Distinctiveness Conflict Risk

Quotes: product-specific triggers like "tu command", "tu build", "mentions installing ToolUniverse or setting up scientific tools". A clear niche (ToolUniverse installation/configuration) with distinct, product-named triggers means minimal conflict risk with other skills — the level-5 anchor; nothing generic like 'helps with tools' that would push it toward level 3-4.

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

relative_links

Relative link issues: 1 missing

Warning

Total

15

/

16

Passed

Repository

mims-harvard/ToolUniverse

Path

plugins/tooluniverse/skills/setup-tooluniverse/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.