skills/devtu-optimize-descriptions/SKILL.md
Optimize tool descriptions in ToolUniverse JSON configs for clarity and usability. Reviews descriptions for missing prerequisites, unexpanded abbreviations, unclear parameters, and missing usage guidance. Use when reviewing tool descriptions, improving API documentation, or when user asks to check if tools are easy to understand.
npx skillsauth add mims-harvard/tooluniverse devtu-optimize-descriptionsInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Optimize tool descriptions in ToolUniverse JSON configuration files to ensure they are clear, complete, and user-friendly.
Use when:
Tool Description Review:
- [ ] Prerequisites stated (packages, API keys, accounts)
- [ ] Critical abbreviations expanded on first use
- [ ] Required vs optional parameters clear
- [ ] Mutually exclusive options numbered/labeled
- [ ] Parameter guidance includes trade-offs
- [ ] Filter syntax shows available fields
- [ ] File size warnings where relevant
- [ ] Examples show realistic usage
Problem: Users don't know if they need ONE input or ALL inputs.
Fix: Use "Required: Provide ONE input type" for mutually exclusive options.
// Before
"description": "Process BED regions, motifs, or gene lists..."
// After
"description": "Process genomic data. **Required: Provide ONE input type** - (1) BED regions, (2) DNA motif, or (3) gene list. Analyzes..."
Number the options and use bold for "Required".
Problem: Users don't know what to install/configure before use.
Fix: Add prerequisites note to first tool in each family.
"description": "Query single-cell data. Prerequisites: Requires 'package-name' (install: pip install tooluniverse[extra]). Returns..."
Include:
Problem: New users don't understand technical terms.
Fix: Expand on first use with format: "Abbreviation (Full Name)".
Common abbreviations to expand:
// Before
"description": "Download H5AD files..."
// After
"description": "Download H5AD (HDF5-based AnnData) files..."
Problem: Users don't know what fields are available or what syntax to use.
Fix: List operators, common fields, and provide multiple examples.
"parameter_name": {
"type": "string",
"description": "Filter using SQL-like syntax. Format: 'field == \"value\"'. Operators: ==, !=, in, <, >, <=, >=. Combine with 'and'/'or'. Common fields: tissue, cell_type, disease, assay, sex, ethnicity. Examples: 'tissue == \"lung\"', 'disease == \"COVID-19\" and tissue == \"lung\"', 'cell_type in [\"T cell\", \"B cell\"]'."
}
Include:
Problem: Users don't know which value to choose or what trade-offs exist.
Fix: Explain what each value means and provide recommendations.
// Before
"threshold": "Q-value threshold (05=1e-5, 10=1e-10, 20=1e-20)"
// After
"threshold": "Peak calling stringency. '05'=1e-5 (permissive, more peaks, broad features), '10'=1e-10 (moderate, balanced), '20'=1e-20 (strict, high confidence, narrow peaks). Default '05' suitable for most analyses. Higher values = fewer but more confident peaks."
For each parameter option, explain:
Problem: Users provide multiple options when only one is allowed.
Fix: Label options as "Option 1", "Option 2", etc.
"bed_data": {
"description": "**Option 1**: BED format regions (tab-separated: chr, start, end). Example: 'chr1\\t1000\\t2000'."
},
"motif": {
"description": "**Option 2**: DNA sequence motif in IUPAC notation. Use: A/T/G/C, W=A|T, S=G|C. Example: 'CANNTG'."
},
"gene_list": {
"description": "**Option 3**: Gene symbols as array. Example: ['TP53', 'MDM2']."
}
For tools that download or return large files:
"description": "Download contact matrices. Note: Files can be large (GBs), check file_size in metadata before downloading. Returns..."
When tool returns submission URL instead of direct results:
"description": "Perform enrichment analysis. Note: Returns submission URL (web form-based analysis). Analyzes..."
For tools with multiple format options:
"file_type": "File format. Common types: 'cooler' (multi-resolution contact matrices), 'pairs' (aligned read pairs), 'hic' (Juicer format), 'mcool' (multi-resolution cooler)."
{
"name": "Tool_operation_name",
"type": "ToolClassName",
"description": "[Action verb] to [purpose]. [Prerequisites if first tool]. [Key data/features]. [Required inputs if mutually exclusive]. [Note about limitations/requirements]. Use for: [use case 1], [use case 2], [use case 3].",
"parameter": {
"properties": {
"param_name": {
"type": "string",
"description": "[What it does]. [Format/syntax if applicable]. [Options with trade-offs]. [Examples]. [Recommendation if applicable]."
}
}
}
}
To verify description quality, ask:
Can a new user understand what the tool does?
Can a user provide correct inputs on first try?
Can a user choose appropriate parameters?
Are prerequisites obvious?
"description": "Query [data type] from [source]. [Prerequisites]. Filter by [criteria]. Returns [output]. [Data scale]. Use for: [discovery], [analysis], [specific research tasks]."
Key elements:
"description": "Download [file types] from [source]. [Format details]. [File size warning]. [Authentication requirement]. Use for: [offline analysis], [custom processing], [integration]."
Key elements:
"description": "Analyze [input type] to find [results]. **Required: Provide ONE input type** - (1) [option], (2) [option], (3) [option]. Compares against [database/background]. [Result format]. Use for: [identifying], [discovering], [predicting]."
Key elements:
After updating descriptions, validate JSON syntax:
# Validate all tool JSONs
python3 -m json.tool src/tooluniverse/data/your_tools.json > /dev/null && echo "✓ Valid"
# Check all tools in category
for f in src/tooluniverse/data/*_tools.json; do
python3 -m json.tool "$f" > /dev/null && echo "✓ $f valid" || echo "✗ $f invalid"
done
Before (Unclear):
{
"name": "Tool_enrichment",
"description": "Perform enrichment with tool to find factors.",
"parameter": {
"properties": {
"bed": {"description": "BED data"},
"motif": {"description": "Motif"},
"genes": {"description": "Genes"},
"threshold": {"description": "Threshold value"}
}
}
}
After (Clear):
{
"name": "Tool_enrichment_analysis",
"description": "Identify transcription factors enriched in your data. **Required: Provide ONE input type** - (1) BED genomic regions, (2) DNA sequence motif (IUPAC notation), or (3) gene symbol list. Compares against 400,000+ ChIP-seq experiments. Returns ranked proteins with enrichment scores. Note: Returns submission URL (web-based analysis). Use for: identifying regulators of regions, finding proteins bound to motifs, discovering transcription factors regulating genes.",
"parameter": {
"properties": {
"bed_data": {
"description": "**Option 1**: BED format regions (tab-separated: chr, start, end). For finding proteins bound to genomic regions. Example: 'chr1\\t1000\\t2000'."
},
"motif": {
"description": "**Option 2**: DNA motif in IUPAC notation (A/T/G/C, W=A|T, S=G|C, M=A|C, K=G|T, R=A|G, Y=C|T). Example: 'CANNTG' (E-box)."
},
"gene_list": {
"description": "**Option 3**: Gene symbols as array or single gene. Example: ['TP53', 'MDM2', 'CDKN1A']."
},
"threshold": {
"description": "Peak stringency. '05'=1e-5 (permissive, more peaks), '10'=1e-10 (moderate), '20'=1e-20 (strict, high confidence). Default '05' suitable for most analyses."
}
}
}
}
Priority order for optimization:
Critical (fix immediately):
High (fix soon):
Medium (nice to have):
Expected impact: 50-75% reduction in user errors, 50-67% faster time to first successful use.
tools
Generate the success criteria for a task or question, then review work against them. Given a task, goal, or open-ended question, decompose it into scenarios, evaluation perspectives, and fine-grained weighted YES/NO criteria using the Recursive Expansion Tree (RET) method; if work is supplied, score it criterion-by-criterion and surface what is missing or could be better. Use when asked to self-review or check your own work, judge whether a task is done well or completely, build a definition-of-done or completeness checklist, create an evaluation rubric or grading criteria, score or grade answers to a question, set up an LLM-as-judge rubric, or when the user mentions self-review, completeness check, success criteria, evaluation criteria, scoring rubric, Qworld, or the RET algorithm.
tools
Find the real protein target(s) of a peptide from its sequence — peptide target deorphanization / off-target identification, for ANY target class (GPCR, ion channel, protease, cytokine/growth-factor receptor, enzyme, integrin), not only GPCRs. Use when a peptide has a phenotype but does not bind its hypothesized target, when a peptide binds a target in one species or assay but not another, or to screen candidate targets for an orphan peptide. A target-class router steers a multi-route keyless pipeline (PROSITE/ELM motif, BLAST homology, HGNC/InterPro/GPCRdb/GtoPdb target-family enumeration, OpenTargets phenotype anchor, EnsemblCompara/Alliance cross-species reconciliation) plus optional NVIDIA-NIM co-folding (Boltz2, AlphaFold2-Multimer, OpenFold3) for structural confirmation.
tools
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.
tools
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 1000+ 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.