Digital Humanity Skills
A Claude Code plugin suite for digital humanities (DH) researchers and students. It guides the full research lifecycle — from research question and corpus building, through writing and peer review, to submission — using structured agent prompts rather than executable code.
Designed for projects anywhere on the DH spectrum: digitized historical archives, literary corpus analysis, cultural analytics, digital editions, and the challenge of connecting computational methods with humanistic argument. Works for anyone from undergraduate to faculty.
What You Get
Five skills, each invoked with a / command in Claude Code:
| Skill | Command | Purpose |
|---|
| dh-explore | /dh-explore | Develop your research question, survey digital archives, assess primary sources, build your corpus, review literature, generate a venue-specific research framework map |
| dh-write | /dh-write | Write proposals, grant applications, thesis chapters, journal articles, book reviews, response-to-reviewers letters, close readings |
| dh-review | /dh-review | Simulate peer review (technical, humanistic, interdisciplinary); parse supervisor feedback into a structured revision plan |
| dh-refs | /dh-refs | Import Zotero/.bib/.ris sources, build annotated bibliographies, format citations, identify literature gaps |
| dh-pipeline | /dh-pipeline | Orchestrate the full 13-stage lifecycle with mandatory integrity gates and a DH Material Passport |
Skills hand off structured artifacts to each other (Research Question Brief → corpus → draft → Reviewer Decision → revision). See docs/ARCHITECTURE.md for the interaction model.
Quick Start
New project — start with discovery:
/dh-explore
Have a research question, need a method + paper structure plan:
/dh-explore framework
Tell it your RQ and target venue (ACL / DHQ / PMLA / specific journal) — it builds a Mermaid diagram of recommended methods and a section-by-section paper structure.
Already have sources — build your bibliography first:
/dh-refs annotate
(paste your Zotero export, .bib file, or reading list)
Ready to write:
/dh-write article
/dh-write thesis
/dh-write grant-proposal
Received reviewer or supervisor feedback:
/dh-review supervisor-feedback
/dh-write response-to-reviewers
(paste the comments into the session)
Dissertation or long-term project — use the full pipeline:
/dh-pipeline
For a worked example and first-run walkthrough, see QUICKSTART.md.
Installation
Prerequisites: Claude Code CLI installed and configured.
Option 1: Plugin install (recommended)
From this repository root:
claude plugin install .
This reads .claude-plugin/plugin.json and registers all five skills.
Option 2: Symlinks (for local development)
mkdir -p ~/.claude/skills
ln -s $(pwd)/dh-explore ~/.claude/skills/dh-explore
ln -s $(pwd)/dh-write ~/.claude/skills/dh-write
ln -s $(pwd)/dh-review ~/.claude/skills/dh-review
ln -s $(pwd)/dh-refs ~/.claude/skills/dh-refs
ln -s $(pwd)/dh-pipeline ~/.claude/skills/dh-pipeline
After installation, open a new Claude Code session and run /dh-explore quick-brief to verify. Full setup and troubleshooting: docs/SETUP.md.
Repository Structure
.
├── .claude-plugin/plugin.json # Plugin manifest — registers all five skills
├── .claude/CLAUDE.md # Routing rules and handoff artifact signatures
├── dh-explore/ # Research & discovery (11 agents, 8 modes)
├── dh-write/ # Writing & drafting (12 agents, 14 modes, 5 templates)
├── dh-review/ # Peer review & feedback parsing (6 agents)
├── dh-refs/ # Citation & reference management (5 agents)
├── dh-pipeline/ # Full lifecycle orchestrator (5 agents)
├── shared/
│ ├── references/ # Cross-skill protocols (ethics, user profiles, intent routing)
│ ├── contracts/passport/ # DH Material Passport JSON schemas
│ └── templates/
├── docs/
│ ├── ARCHITECTURE.md # How skills connect and hand off artifacts
│ └── SETUP.md # Installation and verification
├── MODE_REGISTRY.md # All 34 modes, triggers, and outputs across 5 skills
└── QUICKSTART.md # First-use guide and canonical test project
Each skill directory contains:
SKILL.md — mode routing; dispatches to agent files in sequenceagents/ — self-contained prompt documents for each agent rolereferences/ — skill-specific reference material
There is no build step. Skills are markdown instructions that Claude reads and follows at runtime.
Key Concepts
