A community-maintained catalog of AI agent skills — structured, searchable, and integrated with the OpenModels ecosystem.

Skills are structured definitions of tasks that AI agents excel at. Each skill includes:

• Description — what the skill does and when to use it
• Recommended models — which LLMs work best (linked to OpenModels registry)
• Example prompts — ready-to-use templates
• Metadata — category, complexity, modalities, compatible tools

Skills help developers discover what AI agents can do and choose the right model for each task.

Browse skills on openmodels.run/skills or explore the YAML files directly:

skills/
├── code-review.yaml
├── unit-test-generation.yaml
├── refactoring-assistant.yaml
├── debugging-assistant.yaml
├── api-design.yaml
├── sql-generation.yaml
├── schema-design.yaml
├── documentation-generator.yaml
├── security-audit.yaml
├── accessibility-review.yaml
├── performance-optimization.yaml
├── ci-cd-pipeline.yaml
├── commit-message-writer.yaml
├── prompt-engineering.yaml
├── code-translation.yaml
├── regex-builder.yaml
├── ...
└── 75 skills total (see full list in skills/ directory)

• infrastructure-as-code — Terraform/CDK/Pulumi IaC configuration generation
• threat-modeling — STRIDE analysis and security threat assessment
• video-analysis — video content analysis with scene detection and annotation
• api-load-testing — k6/Locust load testing script generation
• audio-transcription — audio transcription with speaker diarization
• contract-analysis — legal contract review and risk assessment
• incident-response — SRE runbooks and postmortem generation
• mobile-app-prototyping — SwiftUI/Compose/React Native prototyping
• design-system-generator — design tokens and component specifications
• data-pipeline-builder — Airflow/dbt/Spark ETL pipeline generation

• aso-audit — full App Store health check with 10-dimension scored report card (0–100)
• aso-competitor-analysis — keyword gaps, creative strategy, ratings, and monetization benchmarking
• aso-keyword-research — keyword discovery, opportunity scoring, and strategy bucketing
• aso-metadata-optimization — title, subtitle, keyword field, and description writing for iOS and Android

• literature-review — systematic reviews across PubMed, arXiv, bioRxiv
• hypothesis-generation — scientific hypothesis formulation and experimental design
• bioinformatics-pipeline — scRNA-seq, Scanpy, pathway analysis, variant calling
• molecular-analysis — cheminformatics, RDKit, drug-likeness, ADMET
• scientific-visualization — publication-quality figures for journals
• network-graph-analysis — NetworkX, community detection, GNN
• time-series-forecasting — ARIMA, Prophet, LSTM, anomaly detection
• ml-experiment-tracker — MLflow, W&B, Optuna, hyperparameter tuning
• scientific-writing — manuscripts, grants, IMRaD, peer review
• geospatial-analysis — GeoPandas, GIS, remote sensing, spatial stats

Each skill is a YAML file validated against a JSON Schema:

id: code-review
name: Code Review
description: >
 Automated code review with actionable feedback...
category: development
tags:
 - code-quality
 - automation
author:
 name: OpenModels Community
 github: openmodelsrun
recommended_models:
 - claude-opus-4-6
 - gpt-5
 - gemini-2-5-pro
min_context_window: 32000
modalities:
 input: [code, text]
 output: [text, code]
complexity: intermediate
use_cases:
 - Pull request review automation
 - Code quality gates in CI
example_prompt: |
 Review the following code for bugs, performance, and style...
related_skills:
 - unit-test-generation
 - security-audit
compatible_tools:
 - claude-code
 - cursor
 - kiro
 - any
created_at: "2026年05月24日T10:00:00.000Z"
updated_at: "2026年05月24日T10:00:00.000Z"

We welcome contributions! See CONTRIBUTING.md for guidelines.

Quick steps:

1. Fork this repository
2. Create a new YAML file in skills/ following the schema
3. Run validation: python validate.py
4. Submit a pull request

All skills are validated against JSON Schema on every PR:

pip install -r requirements.txt
python validate.py

Validation checks:

• YAML syntax
• Schema conformance (required fields, types, enums, patterns)
• No duplicate IDs
• Filename matches the id field
• Related skills reference existing IDs

Skills are ingested into the OpenModels platform and available via:

• Web UI — browse, filter, and search at openmodels.run/skills
• API — query programmatically via /api/v1/skills
• Cross-references — skills link to models, models link to skills

MIT