Skip to main content
skill Claude Code CX Codex Cursor GitHub Copilot

ncats-arax — Claude Code skill: what it does, how to install

Skill for querying the NCATS Translator ARAX knowledge graph: typed edges with provenance, entity normalization, and offline review of saved TRAPI responses.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add K-Dense-AI/scientific-agent-skills --skill ncats-arax -a claude-code
Open on GitHub

What ncats-arax does

The skill gives the agent one way to question ARAX, the biomedical knowledge graph behind NCATS Translator. You hand it reviewed CURIEs and Biolink types and get relationships back with provenance attached: knowledge sources, publications, the edge bindings behind each result. Two query shapes are supported. A typed one-hop lookup, or exactly two hops with both endpoints pinned.

Everything runs through scripts/arax_client.py, which exposes preflight, normalize, one-hop, two-hop and summarize. The client is standard-library Python, so there is no key to manage. Normalization stays a separate review step: check the proposed CURIE and category before anything reaches the graph. Runs land in summary.json and response.json under an output directory you pick.

You need Python 3.10+ and outbound HTTPS to arax.transltr.io. The part worth reading twice: queries and caller metadata may be publicly visible, so patient data and unpublished target hypotheses do not belong here. Results are capped at 50, and lookup mode defaults to 20. A zero answer means nothing came back under those constraints, not that no relationship exists.

What it helps with

  • find drug-to-target edges in RTX-KG2
  • turn a free-text compound name into a reviewed CURIE
  • trace a two-hop path from molecule to disease with pinned ends
  • pull provenance out of a saved TRAPI response
  • compare two named providers in federated mode

Example prompts for the agent

  • Normalize `ivacaftor` to a CURIE with category `biolink:SmallMolecule` and show the first synonyms.
  • Look up `CHEBI:31690` against `NCBIGene:25` over `biolink:affects` with the decreased qualifier.
  • Summarize `outputs/ivacaftor-cystic-fibrosis/summary.json` and show provenance for each path.

How to install ncats-arax

1Skills CLI (Vercel)

npx skills add K-Dense-AI/scientific-agent-skills --skill ncats-arax -a claude-code

2Manually: put the files in place

~/.claude/skills/ncats-arax/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/ncats-arax/

From the same set120

The whole set →
scientific-visualization K-Dense-AI Generates and audits publication-ready scientific figures in Matplotlib, Seaborn, and Plotly, focusing on truthful data encoding and accessibility. Claude Code CX Codex Cursor +1 2,400installs scientific-writing K-Dense-AI This skill drafts and audits scientific manuscripts and reports while tying claims to verified evidence. Use it for sections, references, declarations, and submission checks. Claude Code CX Codex Cursor +1 2,400installs statistical-analysis K-Dense-AI Runs hypothesis tests with assumption checking, effect sizes, and APA-formatted reports. Covers t-tests, ANOVA, correlation, regression, non-parametric and Bayesian methods. Claude Code CX Codex Cursor +1 2,300installs scientific-brainstorming K-Dense-AI Structures early-stage research ideation through independent idea generation, transparent scoring, and adversarial review. Useful for ranking hypotheses without premature consensus. Claude Code CX Codex Cursor +1 2,300installs literature-review K-Dense-AI This skill searches and synthesizes research literature, checks citations, and prepares Markdown or PDF reviews. Use it for reproducible searches and screening. Claude Code CX Codex Cursor +1 2,300installs scientific-critical-thinking K-Dense-AI Evaluates scientific claims, research methodology, and evidence quality. Applies GRADE and Cochrane Risk of Bias frameworks. Claude Code CX Codex Cursor +1 2,300installs scientific-schematics K-Dense-AI Creates PNG drafts of scientific diagrams and runs a separate quality review. Use it for neural network architectures, flowcharts, system diagrams, and biological pathways. Claude Code CX Codex Cursor +1 2,200installs scikit-learn K-Dense-AI A skill for Python machine learning tasks: classification, regression, clustering, preprocessing, and building ML pipelines with scikit-learn 1.9.1. Claude Code CX Codex Cursor +1 2,200installs

Similar

manimce-best-practices adithya-s-k A skill that steers the agent through Manim Community Edition: scenes, MathTex, 3D, camera work. Useful once a project starts with from manim import *. Claude Code CX Codex Cursor +1 3,900installs ccf-common mikubaka88 Shared preflight module for the CCFA skill suite. It configures routing, working directories, and evidence standards before specialist skills run. Claude Code CX Codex Cursor +1 2,000installs ccf-humanization mikubaka88 A required first pass before every CCFA skill: it strips defensive hedging out of research prose and keeps the evidence, numbers and honest uncertainty. Claude Code CX Codex Cursor +1 1,900installs build-evidence-map official github Builds an auditable evidence map for a disputed decision while retaining support, counterevidence and gaps. Use it when a prose summary would hide disagreement. GitHub Copilot 70installs anysearch anysearch-ai This skill gives an agent web search, domain-specific queries, parallel batch search, and page extraction. Use it when a task needs current information or a claim needs checking. Claude Code CX Codex Cursor +1 38,500installs academic-paper Imbad0202 This skill runs a 12-agent academic pipeline across 11 modes. It guides projects from literature searches and outlines to citation audits and rebuttal reviews. Claude Code CX Codex Cursor +1 10,800installs nature-figure Yuan1z0825 A skill that builds, revises and audits scientific figures for manuscripts in Python or R. Useful when plots go to submission, not to a report. Claude Code CX Codex Cursor +1 18,700installs academic-paper-reviewer Imbad0202 This skill reviews an academic paper through five role-based perspectives. Use it for peer review, revision checks, or a focused methodology assessment. Claude Code CX Codex Cursor +1 9,800installs

Questions about ncats-arax

How do I install ncats-arax in Claude Code?

Run npx skills add K-Dense-AI/scientific-agent-skills --skill ncats-arax -a claude-code, then confirm the file landed at ~/.claude/skills/ncats-arax/SKILL.md.

What does ncats-arax do?

It sends typed queries to the ARAX API and returns graph relationships with provenance. It also normalizes entities to CURIEs and rebuilds a summary from saved files without network access.

Do I need an API key for ncats-arax?

No key, since the client uses only the Python standard library. Python 3.10+ and HTTPS access to arax.transltr.io are required, and queries may be publicly visible, so keep sensitive content out.

Source text

SKILL.md · MIT
---
name: ncats-arax
description: Queries the NCATS Translator ARAX production API for bounded, typed, provenance-rich one-hop and endpoint-pinned two-hop biomedical knowledge-graph relationships. Use for Biolink-constrained RTX-KG2 lookup, explicit selected-provider ARAX federation, separate entity normalization, qualifier-aware graph traversal, and inspection of TRAPI edge bindings, publications, and knowledge-source provenance. Do not use for inference, ranking, open-ended pathfinding, clinical guidance, or sensitive queries.
allowed-tools: Read Bash
license: MIT
compatibility: Requires Python 3.10+ and outbound HTTPS access to arax.transltr.io. The client uses only the Python standard library and needs no API key. Queries and caller metadata may be publicly visible; never submit sensitive or patient-specific content.
metadata:
  version: "1.2"
  last-reviewed: "2026-09-30"
  skill-author: neuroepithelial
---

# NCATS ARAX

Use ARAX as a constrained knowledge-graph lookup service. Submit reviewed CURIEs and explicit
Biolink types, preserve the exact TRAPI exchange, inspect query-edge bindings and provenance, and
treat every returned path as a candidate for subsequent verification.

Read [query-contract.md](references/query-contract.md) before constructing a query. Read
[output-schema.md](references/output-schema.md) when interpreting saved artifacts, warnings,
provenance, or partial results.

## Safety boundary

- Use only public, nonsensitive research questions. ARAX status facilities may expose query and
  caller metadata even when `store=false` is requested.
- Do not submit patient information, confidential research questions, unpublished compound
  programs, or proprietary target hypotheses.
- Do not present a returned path as a validated mechanism or clinical recommendation.
- Report a zero as "not returned under these constraints," never as evidence that no relationship
  exists.
- Describe position as unscored response order, never rank.
- Verify important candidates with literature and authoritative databases separately.

## Workflow

1. Normalize free text separately, then review and report the proposed CURIE and category.
2. Choose a typed one-hop query or an exactly two-hop query with both endpoints pinned.
3. Use default RTX-KG2 lookup unless the user explicitly names two to five providers.
4. Acknowledge that the biomedical query is public and choose a new or empty output directory.
5. Run the client once. Do not silently change provider selection or expansion order after a
   failure or empty result.
6. Inspect `summary.json` for bounded bindings and provenance and `response.json` for the exact
   TRAPI payload.
7. Verify scientifically important paths outside ARAX.

## Preflight

Check the production OpenAPI without making a biomedical query:

```bash
python skills/ncats-arax/scripts/arax_client.py preflight
```

The client verifies that the service identifies itself as ARAX, exposes `POST /query` and
`GET /entity`, and reports a supported TRAPI version. It reads `info.x-trapi.version`, falling back
to the title for older OpenAPI documents. A nonproduction endpoint or untested TRAPI series
requires an explicit override; neither override changes the fixed query shapes or operations.

## Normalize an entity

Normalization is review-only and never triggers a graph query:

```bash
python skills/ncats-arax/scripts/arax_client.py normalize "ivacaftor" \
  --expected-category biolink:SmallMolecule \
  --max-synonyms 10 \
  --acknowledge-public-query \
  --output-dir outputs/normalize-ivacaftor
```

Review the canonical identifier, name, category, and synonym preview before using a CURIE. Report
all CURIEs and categories regardless of query outcome. A category warning or zero result is a
reason to curate the identifier, not to chain automatically to `/query`.

## One-hop lookup

Pin at least one endpoint and type both nodes:

```bash
python skills/ncats-arax/scripts/arax_client.py one-hop \
  --subject-id CHEBI:31690 \
  --subject-category biolink:SmallMolecule \
  --predicate biolink:affects \
  --object-id NCBIGene:25 \
  --object-category biolink:Gene \
  --qualifier biolink:object_aspect_qualifier=activity_or_abundance \
  --qualifier biolink:object_direction_qualifier=decreased \
  --acknowledge-public-query \
  --output-dir outputs/imatinib-abl1
```

Lookup mode is the default and fixes expansion to `infores:rtx-kg2`. It defaults to 20 results.
Use `--result-limit N` to request 1-50 results; 50 is the hard cap in either mode.

## Endpoint-pinned two-hop lookup

Use exactly one typed, unpinned intermediate node:

```bash
python skills/ncats-arax/scripts/arax_client.py two-hop \
  --subject-id CHEBI:66901 \
  --subject-category biolink:SmallMolecule \
  --predicate-1 biolink:affects \
  --intermediate-category biolink:Gene \
  --predicate-2 biolink:associated_with \
  --object-id MONDO:0009061 \
  --object-category biolink:Disease \
  --qualifier-1 biolink:object_aspect_qualifier=activity_or_abundance \
  --qualifier-1 biolink:object_direction_qualifier=increased \
  --expand-order right-first \
  --acknowledge-public-query \
  --output-dir outputs/ivacaftor-cystic-fibrosis
```

Right-first expansion is the default. If an empty result merits another attempt, run a new query
explicitly with `--expand-order left-first` and keep the runs separate.

## Selected-provider federation

Federation is explicit and accepts two to five named providers:

```bash
python skills/ncats-arax/scripts/arax_client.py one-hop \
  --subject-id CHEBI:31690 \
  --subject-category biolink:SmallMolecule \
  --predicate biolink:affects \
  --object-id NCBIGene:25 \
  --object-category biolink:Gene \
  --mode federated \
  --kp infores:rtx-kg2 \
  --kp infores:molepro \
  --acknowledge-public-query \
  --output-dir outputs/federated-imatinib-abl1
```

Federation defaults to the hard maximum of 50 results. Provider errors may coexist with useful
results; such a run exits 7 after retaining its artifacts and is marked partial. The same applies
to a failed provider in lookup mode. An explicit non-success ARAX response status exits 6 with
the raw response retained; it must not be reported as a successful zero-result query.

## Inspect saved provenance

Rebuild a bounded summary without network access:

```bash
python skills/ncats-arax/scripts/arax_client.py summarize \
  --request outputs/ivacaftor-cystic-fibrosis/request.json \
  --response outputs/ivacaftor-cystic-fibrosis/response.json \
  --format text
```

The inspector accepts only the same constrained request shapes and fixed operations that the live
commands generate. Use `--format json` for the normalized view on standard output.

## Interpret results

- Follow each analysis's query-edge bindings; do not summarize every knowledge-graph edge.
- Preserve the physical edge subject, predicate, object, and qualifier values returned by ARAX.
  Returned predicates or qualifier aspects may be more specific than the query constraint.
- Inspect all source objects, including primary, aggregator, supporting-data, upstream-resource,
  and source-record URL fields.
- Trace aggregator edges to their primary/upstream sources and publications before claiming
  corroboration. Multiple providers can redistribute the same record; report distinct primary
  evidence, not provider count as confidence or independent replication.
- Treat `publication_availability: not_returned` as missing metadata, not evidence that no
  publications exist.
- Treat missing auxiliary-graph references and provider failures as explicit warnings.
- Consult the raw response whenever the bounded summary omits detail or the service response is
  partial, unfamiliar, or scientifically surprising.

## Deliberate exclusions

The client has no raw-query, workflow, operation, overlay, ranking, inference, link-prediction,
Pathfinder, ARS, batch, all-provider, three-hop, cache, daemon, SDK, MCP, or
natural-language-to-TRAPI surface. Do not work around those limits with direct HTTP calls under
this skill.

## Official references

Reviewed on 2026-09-30 against production ARAX 1.5.4 / TRAPI 1.5.0 (the URL still contains `v1.4`).
Live public smoke tests passed for preflight, normalization, qualified one-hop and endpoint-pinned
two-hop lookups, and RTX-KG2/MolePro federation. Results and provider availability can change.
The official introductory guide contains older response examples; use the deployed schema and
current ARAX source for field and operation contracts. No Python SDK is used by this client.

- [ARAX documentation](https://ncatstranslator.github.io/TranslatorTechnicalDocumentation/architecture/ara/arax/)
- [ARAX production OpenAPI](https://arax.transltr.io/api/arax/v1.4/openapi.json)
- [ARAXi operation documentation](https://github.com/RTXteam/RTX/blob/master/code/ARAX/Documentation/DSL_Documentation.md)
- [ARAX query execution and scoreless resultification](https://github.com/RTXteam/RTX/blob/master/code/ARAX/ARAXQuery/ARAX_query.py)
- [Translator Reasoner API](https://github.com/NCATSTranslator/ReasonerAPI)
- [Biolink Model](https://biolink.github.io/biolink-model/)

© K-Dense-AI · MIT · K-Dense-AI/scientific-agent-skills

See also

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor