Skip to main content
skill CX Codex

tooluniverse-aging-senescence — Claude Code skill: how to install

tooluniverse-aging-senescence structures agent reasoning on aging biology, senescence pathways, and senolytics. It prioritizes causal evidence over correlation across four tiers.

Apache-2.0 230 installs ★ 1,798 updated 2026-10-24

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-aging-senescence -a claude-code
Open on GitHub

What tooluniverse-aging-senescence does

The skill organizes research queries around the 12 hallmarks of aging framework. Rather than guessing facts from memory, the agent routes lookups through databases like Open Genes, GWAS Catalog, and PubMed. Queries can call AgingCohort_search to verify longitudinal cohorts such as EPESE or ALSA, while OpenGenes_get_gene returns mechanism summaries and experimental counts.

Evidence is ranked systematically: T1 for human genetics and centenarian cohorts, T2 for model organisms, T3 for cell culture assays, and T4 for computational predictions. The prompt strictly keeps in vitro T3 observations separated from T1 clinical proof.

Because key FOXO3 longevity trials used targeted genotyping, they are absent from the GWAS Catalog. The workflow explicitly accounts for this gap by adding targeted literature searches.

What it helps with

  • classify aging-related genes across the 12 hallmarks framework
  • screen senolytic drug candidates for age-related pathologies
  • evaluate cellular senescence markers including p16 and SASP factors
  • distinguish correlative markers from causal intervention data

Example prompts for the agent

  • What genes are associated with longevity and what evidence tier supports FOXO3?
  • Find senolytic drug candidates targeting senescent cells in osteoarthritis
  • Map ATM and TP53 to aging hallmarks and check their SASP interactions

How to install tooluniverse-aging-senescence

1Skills CLI (Vercel)

npx skills add mims-harvard/ToolUniverse --skill tooluniverse-aging-senescence -a claude-code

2Manually: put the files in place

~/.claude/skills/tooluniverse-aging-senescence/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/tooluniverse-aging-senescence/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about tooluniverse-aging-senescence

How to install tooluniverse-aging-senescence in Claude Code?

Run npx skills add mims-harvard/ToolUniverse --skill tooluniverse-aging-senescence -a claude-code. The skill file is placed in ~/.claude/skills/tooluniverse-aging-senescence/SKILL.md.

What does tooluniverse-aging-senescence do?

It structures biomedical queries on aging mechanisms, cellular senescence, and senolytic target discovery using a phased workflow.

Why does the skill differentiate evidence tiers?

It separates T1 human genetics from T2 animal data, T3 in vitro senescence, and T4 predictions, ensuring cell culture correlations are not confused with causal human longevity drivers.

Source text

SKILL.md · Apache-2.0
---
name: tooluniverse-aging-senescence
description: Aging biology, cellular senescence, and longevity research. Covers senescence markers (p16/CDKN2A, SASP, SA-beta-gal), aging hallmarks, senolytic drug discovery (dasatinib+quercetin, fisetin, navitoclax), epigenetic clocks, telomere biology, and longevity GWAS. Use for senescence-pathway analysis, age-related disease genetics, senolytic-target discovery, and centenarian-genetics queries. Distinguishes correlative vs causal evidence (knockout, intervention).
disable-model-invocation: true
---

# Aging & Cellular Senescence Research

## Aging Research Reasoning

Before querying any tool, ask the central question: is this a cause or consequence of aging?

Senescence markers (SA-β-gal, p16/CDKN2A, SASP factors like IL-6 and IL-8) indicate that senescent cells are present. But their presence does not prove that senescence is driving the phenotype. Correlation is easy to establish. Causation requires an intervention. If senolytic drugs (dasatinib+quercetin, fisetin, navitoclax) clear senescent cells and the age-related phenotype improves, that is causal evidence. If clearing senescent cells has no effect, something else is driving the pathology.

Apply this reasoning when interpreting any gene or pathway query: classify it first by hallmark, then ask whether the evidence for its role is correlative (expression data, GWAS association) or causal (functional assay, genetic knockout, senolytic intervention).

Evidence grade the findings: T1 is human genetic evidence (GWAS, centenarian studies). T2 is model organism lifespan data. T3 is cell culture senescence data. T4 is computational prediction. Do not conflate T3 cell culture data with T1 human evidence — they are very different levels of confidence.

A final principle: cellular senescence is one hallmark of aging, not aging itself. Distinguish senescence from organismal aging, from age-related disease, and from progeria (accelerated aging syndromes). These require different tools and different interpretations.

## LOOK UP, DON'T GUESS
When uncertain about any scientific fact, SEARCH databases first (PubMed, UniProt, ChEMBL, ClinVar, etc.) rather than reasoning from memory. A database-verified answer is always more reliable than a guess.

## When to Use

- "What genes are associated with longevity?"
- "Find senolytic drug candidates for [disease]"
- "What are the markers of cellular senescence?"
- "How does [gene] relate to aging?"
- "GWAS hits for age-related diseases"
- "Pathways involved in cellular senescence"
- "What drugs target senescent cells?"

**Not this skill**: For rare disease genetics, use `tooluniverse-rare-disease-diagnosis`. For general disease research, use `tooluniverse-disease-research`.

---

## Workflow

```
Phase 0: Query Parsing — aging gene, senescence marker, age-related disease, or drug query
    |
Phase 1: Hallmarks Classification — map to the 12 hallmarks of aging framework
    |
Phase 2: Genetic Evidence — GWAS, longevity loci, model organism data
    |
Phase 3: Pathway Analysis — senescence, autophagy, telomere, epigenetic pathways
    |
Phase 4: Senolytic/Geroprotector Drug Discovery — existing drugs, clinical trials
    |
Phase 5: Literature & Clinical Context — published evidence, ongoing trials
    |
Phase 6: Interpretation & Report — evidence-graded findings with translational potential
```

---

## Phase 1: Hallmarks Classification

Organize findings around the 12 hallmarks of aging (Lopez-Otin et al., Cell 2023). When a user asks about an aging gene, first classify which hallmark(s) it belongs to, then investigate that hallmark's pathway and disease connections. This prevents scattershot querying — each hallmark has specific pathways and tool strategies.

The hallmarks most amenable to ToolUniverse investigation are: genomic instability (DNA repair genes: ATM, ATR, BRCA1/2, TP53), telomere attrition (TERT, TERC, POT1), epigenetic alterations (DNMT1/3, TET1-3, SIRT1-7), loss of proteostasis (autophagy pathway hsa04140), deregulated nutrient sensing (mTOR pathway hsa04150, FOXO pathway hsa04068, AMPK, IGF1), mitochondrial dysfunction (PINK1, PARKIN, PGC1α), and cellular senescence (CDKN2A/p16, CDKN1A/p21, TP53, RB — KEGG pathway hsa04218).

For altered intercellular communication, focus on SASP factors: IL6, IL8, MCP1 (CCL2), MMP3, MMP9, PAI1, IGFBP7, VEGF. These are the secreted signals that make senescent cells pathological for surrounding tissue.

---

## Phase 2: Genetic Evidence

The best human evidence for aging genes comes from longevity GWAS and centenarian studies. Well-established loci include: APOE (19q13.32, strongest longevity signal), FOXO3 (5q33.3, replicated across multiple centenarian cohorts), TERT (10q24, telomere length GWAS), and CDKN2A/B (9p21.3, GWAS for CVD, cancer, and T2D — all age-related diseases sharing this locus).

Important caveat: many FOXO3 longevity studies (Willcox 2008, Flachsbart 2009) used targeted genotyping rather than GWAS arrays, so they do not appear in the GWAS Catalog. Always supplement GWAS Catalog queries with PubMed literature searches for centenarian studies.

**Finding the underlying cohort a study drew on**: `AgingCohort_search` looks up named longitudinal aging-research cohorts (e.g. EPESE — Established Populations for Epidemiologic Studies of the Elderly; ALSA — Australian Longitudinal Study of Ageing) by keyword/country, returning the cohort's full name, country, and description (verified live, no auth required). Use this when a paper cites a cohort by acronym and you need to know what population/design it represents before judging the evidence tier, or when scoping a new aging analysis and want to know what comparable longitudinal cohorts already exist.

**Start gene-centric questions with Open Genes** — a manually-curated aging-gene database that already aggregates the experimental evidence per gene (lifespan-change studies, longevity associations, age-related expression changes, progeria associations) plus the aging mechanism(s) and functional cluster(s). `OpenGenes_get_gene(symbol="FOXO3")` returns the aging mechanisms, curator confidence level, and an `evidence_counts` summary (e.g. FOXO3 → 58 longevity-association studies) — use it to triage whether a gene is an established aging gene and by what mechanism before drilling into GWAS/OpenTargets/PubMed. `OpenGenes_search_genes` browses the full catalog (~2400 genes). It catches the targeted-genotyping FOXO3 evidence that the GWAS Catalog misses.

```python
# Aging-evidence triage FIRST — mechanisms + curated study counts per gene
OpenGenes_get_gene(symbol="FOXO3")

# Best for gene-centric GWAS analysis
gwas_get_snps_for_gene(gene_symbol="FOXO3")

# For trait queries — note "longevity" is not a standard EFO term; try "lifespan" or specific diseases
gwas_search_associations(query="telomere length")

# OpenTargets aggregated evidence
OpenTargets_get_associated_targets_by_disease_efoId(efoId="EFO_0004847", limit=20)

# Essential for centenarian studies not in GWAS Catalog
PubMed_search_articles(query="FOXO3 GWAS longevity centenarian meta-analysis")
```

---

## Phase 3: Pathway Analysis

The central senescence pathway is KEGG hsa04218. Start there when investigating any senescence-related gene. Supporting pathways: autophagy (hsa04140, implicated in senescence clearance and proteostasis), mTOR signaling (hsa04150, rapamycin target), FOXO signaling (hsa04068, stress resistance and autophagy), and p53 signaling (hsa04115, DNA damage response).

```python
KEGG_get_pathway_genes(pathway_id="hsa04218")   # Cellular senescence
kegg_search_pathway(keyword="autophagy")          # hsa04140
kegg_search_pathway(keyword="mTOR signaling")     # hsa04150
kegg_search_pathway(keyword="FOXO signaling")     # hsa04068
kegg_search_pathway(keyword="p53 signaling")      # hsa04115
```

For SASP network analysis, STRING and Reactome are the right tools:

```python
sasp_genes = ["IL6", "IL8", "MCP1", "MMP3", "MMP9", "PAI1", "IGFBP7", "VEGF", "CCL2"]
STRING_get_network(identifiers="\r".join(sasp_genes), species=9606)
ReactomeAnalysis_pathway_enrichment(identifiers=" ".join(sasp_genes))
```

### Interpreting Senescence Markers

Markers must be interpreted together, not individually. p16 (CDKN2A) upregulation is the closest to a gold standard — it marks irreversible cell cycle arrest — but it is also elevated in some cancers. p21 (CDKN1A) can reflect either transient quiescence or permanent senescence, so it is not specific. SA-β-gal is a lysosomal activity assay that can give false positives in high-confluence cultures. SASP factors (IL-6, IL-8) are also elevated in infection and autoimmunity. γH2AX foci are transient in normal DNA damage but persistent in senescence. Telomere shortening is only relevant for replicative senescence, not for oncogene-induced senescence.

Use a panel. A cell with p16↑ + SA-β-gal↑ + SASP↑ + γH2AX↑ is senescent. A cell with only one marker may not be.

---

## Phase 4: Senolytic and Geroprotector Drug Discovery

Senolytics selectively kill senescent cells. The most clinically advanced combination is dasatinib + quercetin (D+Q), currently in Phase II trials for idiopathic pulmonary fibrosis and diabetic kidney disease. Navitoclax (BCL-2/BCL-XL inhibitor) has strong preclinical data but causes thrombocytopenia, limiting clinical use. Fisetin has Phase II trials for frailty. UBX0101 failed Phase II for osteoarthritis.

Geroprotectors slow aging rather than removing senescent cells. Rapamycin (mTOR inhibitor) extends mouse lifespan and is FDA-approved for transplant. Metformin (AMPK activator) is being tested in the TAME trial. NAD+ precursors (NMN, NR) are in Phase II trials.

```python
DGIdb_get_drug_gene_interactions(genes=["BCL2", "BCL2L1", "TP53", "CDKN2A"])
search_clinical_trials(condition="senescence", query_term="senolytic")
search_clinical_trials(condition="aging", query_term="dasatinib quercetin")
ChEMBL_search_drugs(query="navitoclax")
```

When evaluating a drug candidate, always check clinical status: preclinical data in mice does not translate reliably to humans (telomere biology differs substantially between species). Prioritize T1 human evidence.

---

## Phase 5: Literature and Clinical Context

```python
PubMed_search_articles(query="cellular senescence senolytics clinical trial", max_results=20)
search_clinical_trials(condition="cellular senescence")
search_clinical_trials(query_term="rapamycin aging")
```

---

## Phase 6: Report Structure

1. **Hallmarks Classification** — which hallmarks are relevant and why
2. **Genetic Evidence** — GWAS loci, longevity genes, evidence grade (T1-T4)
3. **Pathway Analysis** — relevant pathways with key genes
4. **Senescence Markers** — expression evidence with interpretation caveats
5. **Drug Candidates** — senolytics and geroprotectors with evidence grade and clinical status
6. **Clinical Trials** — ongoing trials
7. **Mechanistic Model** — how the gene/pathway contributes to aging (cause or consequence?)
8. **Research Gaps** — what interventional data would resolve the causal question

---

## Age-Dependent Expression Analysis

GTEx provides tissue-level median expression but not directly age-stratified data. For age-dependent expression analysis, search PubMed for published GTEx age studies, or use GEO datasets with age metadata.

```python
# GTEx tissue expression (not age-stratified directly)
GTEx_get_median_gene_expression(gene_symbol="CDKN2A")

# Search for published age-expression analyses
PubMed_search_articles(query="GTEx age-dependent expression CDKN2A")
```

---

## Limitations

- Aging is multifactorial — no single gene or pathway explains it; this skill investigates specific aspects
- Mouse lifespan data does not reliably translate to humans (different telomere biology, metabolic rate)
- No single senescence marker is definitive; use a panel (p16 + SA-β-gal + SASP + γH2AX)
- No FDA-approved senolytic exists yet; most trials are Phase I/II
- Epigenetic clocks (Horvath/Hannum) require methylation array data processing not directly queryable via ToolUniverse

© mims-harvard · Apache-2.0 · mims-harvard/ToolUniverse

Similar skills

All skills for Science and research →

From the same set136

The whole set →
tooluniverse-regulatory-variant-analysis tooluniverse-regulatory-variant-analysis interprets non-coding and regulatory genomic variants using GWAS Catalog, GTEx eQTLs, ENCODE chromatin data, and AlphaGenome predictions. CX Codex +1 232installs tooluniverse-computational-biophysics tooluniverse-computational-biophysics works through quantitative biology and biophysics problems, from drug clearance calculations to population-genetics comparisons. CX Codex +1 232installs tooluniverse-rare-disease-genomics tooluniverse-rare-disease-genomics is a Claude Code skill for rare-disease research, tracing patient phenotypes to candidate genes and variants and finding clinical trials. CX Codex +1 231installs tooluniverse-cell-line-profiling The tooluniverse-cell-line-profiling skill helps you pick the right cancer cell line for research by matching genes, tumor type, and drug sensitivity across DepMap, Cellosaurus, COSMIC, and PharmacoDB. CX Codex +1 230installs tooluniverse-vaccine-design tooluniverse-vaccine-design designs computational peptide vaccine candidates using IEDB epitope prediction and HLA coverage. Fits automated workflows for immunogenicity screening. CX Codex +1 229installs tooluniverse-variant-functional-annotation tooluniverse-variant-functional-annotation annotates protein variants with ProtVar, ClinVar, gnomAD, CADD, and ClinGen data. Useful for functional context, population evidence, and variant review. CX Codex +1 229installs tooluniverse-stem-cell-organoid tooluniverse-stem-cell-organoid guides stem cell protocols, pluripotency verification, and organoid fidelity scoring using ToolUniverse biomedical tools. CX Codex +1 226installs tooluniverse-noncoding-rna tooluniverse-noncoding-rna guides non-coding RNA queries across miRBase, LNCipedia, and RNAcentral to inspect targets, transcript structures, and disease links. CX Codex +1 225installs

Project sponsors

Companies that keep this analytics open