Skip to content
TakoAPI
Back to skills

phenoskill

by kaichop

Coding Agents & IDEs
388Downloads reported by clawskills.sh
1Views
0 likesSign in to like this skill
—Not yet rated
Your rating

Brief

Extract clinical phenotypes and medication entities from user-provided text using PhenoSnap, producing.

Description

What This Skill Does

Extracts clinical phenotypes (mapped to HPO terms) and medication entities from user-supplied text using PhenoSnap. Runs locally, redacts identifiable information before writing anything to disk, and saves results as a timestamped JSON file. Does not upload user text anywhere.

Handles sensitive clinical text without sending it to any external service, with built-in PHI redaction before any file is written.

When to Use It

  • Mapping a personal symptom list to HPO phenotype codes
  • Extracting structured medication data from a clinical note
  • Parsing dosage and frequency patterns from a medication list
  • Building a local phenotype record from free-text patient history
  • Structuring symptoms and drugs from a rare disease case description

Example Workflow

INPUT

User asks: Symptoms: developmental delay, seizures, ataxia. Meds: valproate 250 mg BID.

AGENT
  1. 1

    Scans input for identifying information and redacts emails, phone numbers, addresses, and name fields

  2. 2

    Checks whether PhenoSnap is installed locally and downloads it from GitHub if missing

  3. 3

    Verifies the HPO OBO file exists at the configured path

  4. 4

    Writes redacted input to a timestamped text file

  5. 5

    Runs extract_phenotypes.py against the input and HPO OBO file, saving output as JSON

OUTPUT

Timestamped JSON file containing extracted HPO phenotype terms and medication entities, stored locally

Requirements

  • hp.obo file from the Human Phenotype Ontology project (not downloaded automatically by the skill)
  • HPO_OBO_PATH environment variable set to the hp.obo file path, or hp.obo placed manually at resources/hp.obo

Installation

clawhub install kaichop/phenoskill

Run the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.

Links