ovachiever/droid-tings — Summary

droid-tings is a curated collection of 375 skills and 155 custom "droids" (Factory-compatible agents) aggregated from 15+ open-source repositories, targeting both Claude Code and Factory's Droid system. The collection covers AI/ML (80+ skills including LLM frameworks, training tools, inference engines), scientific research (120+ skills including bioinformatics, cheminformatics, protein structure databases), development/engineering (100+ skills including 18 Cloudflare Workers skills), business/product (40+ skills), and utilities/automation (30+ skills). Droids use persona-md format with Claude Code frontmatter (model: claude-opus-4-1-20250805). Two custom commands (start and understand) provide project onboarding and UFIO (Understand First, Implement Once) methodology respectively. The GenerateDroid tool reference in agent frontmatter suggests integration with Factory's dynamic agent creation system. Compared to seeds, droid-tings is purely an aggregation repository — it curates from 15+ upstream sources rather than authoring original content, similar to voltagent-subagents in structure but targeting Factory's Droid platform as the primary runtime alongside Claude Code.

ovachiever/droid-tings — Overview

Origin

Created by ovachiever (44 stars, last push 2025-11-25). Aggregation project pulling from 15+ open-source repositories.

Philosophy

From README:

"A comprehensive, curated collection of 375 skills and 155 custom droids for Claude AI and Factory's Droid system. This repository aggregates the best skills and droids from multiple sources into a single, well-organized collection ready for immediate use."

Dual Target Platform

Unlike other repos in this batch, droid-tings explicitly targets two platforms:

1. Claude Code — standard .claude/agents/ format
2. Factory's Droid system — .factory/skills/ and .factory/agents/ format

Sources

Aggregated from 15+ repositories including (implied from skill diversity): VoltAgent, various scientific computing repos, Cloudflare skill collections, and original custom droids.

Notable Verticals

• Scientific Research (120+ skills): Unusually deep scientific computing coverage — bioinformatics (scanpy, scvi-tools, AnnData), cheminformatics (RDKit, DeepChem), protein databases (PDB, UniProt, AlphaFold), clinical (clinical reports, treatment plans)
• AI/ML (80+ skills): LLM frameworks (LangChain, LlamaIndex, DSPy), training tools (TRL, Axolotl, DeepSpeed), inference engines (vLLM, TensorRT-LLM)
• Quality & Regulatory: ISO 13485, ISO 27001, FDA, GDPR, MDR specialists

UFIO Methodology

The understand command implements "Understand First, Implement Once" — a requirements-before-coding methodology with multi-solution generation and approval gates before implementation.

ovachiever/droid-tings — Architecture

Distribution

Clone and copy.

git clone https://github.com/ovachiever/droid-tings.git
cd droid-tings

# For Claude Code
cp -r skills/* ~/.claude/skills/
cp -r droids/* ~/.claude/agents/

# For Factory Droid
cp -r skills/* .factory/skills/
cp -r droids/* .factory/agents/

Directory Structure

droids/                 # 155 persona-md agent files
  ai-engineer.md
  backend-architect.md
  python-developer.md
  ...
skills/                 # 375 skill folders
  accelerate/
  aeon/
  agentdb-vector-search/
  agile-product-owner/
  ai-sdk-core/
  alphafold-database/
  ...
commands/               # 2 slash commands
  start.md              # Project onboarding briefing
  understand.md         # UFIO methodology

Target AI Tools

• Claude Code (primary: .claude/agents/)
• Factory's Droid system (.factory/skills/, .factory/agents/)

Required Runtime

None beyond Claude Code or Factory.

No Build System

Pure file aggregation. No adapters, no Makefile, no generate scripts.

ovachiever/droid-tings — Components

Totals

Skills (375, domain breakdown)

Sample Skills

AI/ML: LangChain, LlamaIndex, DSPy, Guidance, Instructor, TRL, Axolotl, Unsloth, DeepSpeed, vLLM, SGLang, TensorRT-LLM, LoRA, QLoRA, LlamaGuard, NeMo Guardrails, Constitutional AI, MLflow, Weights & Biases

Scientific: Biopython, scanpy, scvi-tools, PyDESeq2, AnnData, RDKit, DeepChem, AlphaFold, PubMed, UniProt, PDB, ChEMBL, KEGG, DrugBank

Development: React, Next.js, TailwindCSS, shadcn/ui, Zustand, Node.js, FastAPI, Django, PostgreSQL, Drizzle ORM, Supabase, Cloudflare Workers (18 skills), Playwright, Vitest, Jest

Business: PRD templates, roadmaps, user stories, content creation, SEO, ISO 13485, ISO 27001, FDA compliance, GDPR, clinical reports

Droids (155, representative)

• ai-engineer — LLM apps, RAG, vector search, agent orchestration
• backend-architect — server-side architecture
• blue-team-specialist — defensive security
• blockchain-developer — Web3 development
• business-analyst — requirements and analysis
• cloud-architect — cloud infrastructure
• code-reviewer — code quality review
• content-marketer — content creation
• python-developer — (not found, but similar agents present)
• agents-md-documentation-generator — AGENTS.md generation
• agents-md-generator — agent documentation generation
• comment-canonicalizer — code comment standardization
• compliance-style-gate — compliance enforcement

Commands (2)

start — Project Onboarding Briefing

Systematic analysis using project-onboarding-briefer droid:

• Complete system overview + architecture layers + data flow visualization
• Git evolution analysis, code statistics, quality validation
• Professional assessment with ratings + enhancement recommendations
• Phased onboarding roadmap for new team members

understand — UFIO (Understand First, Implement Once)

• Requirements extraction and validation before coding
• Multi-solution approach generation with trade-off analysis
• Detailed implementation planning with approval gates
• Single-session execution for first-time success
• Comprehensive verification and rollback procedures

ai-engineer Droid (Verbatim)

---
name: ai-engineer
description: Build LLM applications, RAG systems, and prompt pipelines. Implements vector search, agent orchestration, and AI API integrations. Use PROACTIVELY for LLM features, chatbots, or AI-powered applications.
model: claude-opus-4-1-20250805
tools: ["Read", "LS", "Grep", "Glob", "Create", "Edit", "MultiEdit", "Execute", "WebSearch", "FetchUrl", "TodoWrite", "Task", "GenerateDroid"]
---

Note: GenerateDroid in tools list is a Factory-specific tool for dynamic droid creation.

ovachiever/droid-tings — Uniqueness & Positioning

Differs From Seeds

Pure aggregation repo targeting two platforms (Claude Code + Factory's Droid system). No seed targets Factory. The scientific research depth (120+ skills covering bioinformatics, cheminformatics, protein structure databases) has no equivalent in seeds. The GenerateDroid tool in agent frontmatter is a Factory-specific capability enabling dynamic agent creation — a runtime-native feature absent from all seeds. The UFIO (Understand First, Implement Once) understand command is one of the few explicit approval-gate patterns in this batch.

Distinctive Position

Scientific/research vertical depth — 120+ skills for scientific computing tasks is unique in this batch and broader corpus. The dual-target (Claude Code + Factory) architecture without a code transformation layer suggests these formats are close enough to be compatible.

Observable Failure Modes

1. Aggregation without provenance: Skills aggregated from 15+ repos may have conflicting conventions, quality, and update cadence.
2. GenerateDroid in Claude Code: Droids with GenerateDroid in their tools list will fail silently in Claude Code (tool not available) but work in Factory.
3. Model version pinning: claude-opus-4-1-20250805 will become an invalid model name as Anthropic deprecates specific versions.
4. Low activity: 44 stars, last push 2025-11-25 — possibly unmaintained.