con/serve

Conserve and serve digital research artifacts.

A comprehensive knowledge base cataloging tools, approaches, and infrastructure for archiving ALL digital research artifacts into git/git-annex/DataLad repositories. Extends YODA and STAMPED principles beyond code and data to encompass communications (Slack, Telegram, Matrix), media (YouTube, Zoom), code artifacts (GitHub issues, PRs, discussions), AI coding sessions, publications, and self-hosted infrastructure.

The name "con/serve" captures both conservation (preserving digital artifacts before they become frozen/inaccessible) and serving (making archived knowledge available to humans and AI systems).

Quick Start

Prerequisites

• uv (Python package manager)
• Git with git-annex

Build and serve locally

# Set up environment and install Hugo
uv venv && source .venv/bin/activate && uv pip install hugo

# Initialize theme submodule
git submodule update --init

# Serve locally with live reload
hugo server

Then open http://localhost:1313/ in your browser.

Build for production

hugo --minify

Output goes to public/.

Content Structure

content/
  _index.md                      # Homepage with architecture diagram
  tools/
    communications/              # Slack, Telegram, Matrix, Mattermost
    media/                       # YouTube (annextube), video, images
    code-artifacts/              # Issues, PRs, discussions, wikis
    cloud-storage/               # rclone and cloud providers
    publications/                # Citations, references, PDFs
    web/                         # Web page and site archival
    ai-sessions/                 # AI coding session capture
  infrastructure/                # Self-hosted services (Forgejo, HedgeDoc, etc.)
  concepts/                      # Cross-cutting patterns and principles
  about/                         # Vision, contributing guide

Taxonomy System

Every tool is classified along four axes:

Axis	Values	Purpose
Category	Communications, Media, Code Artifacts, Web, Cloud Storage, Publications, AI Sessions, Infrastructure	What kind of artifact
Integration	native-datalad, git-annex, git-only, external	How deeply it integrates with the git-annex/DataLad stack
AI Readiness	ai-ready, ai-partial, ai-manual	How easily AI systems can consume the archived output
Media Type	slack, telegram, youtube, github-issues, etc.	Specific platform or format

Adding a New Tool

See Contributing for the full guide. In brief:

1. Create a new .md file in the appropriate content/tools/<category>/ directory
2. Fill in the front matter template with all required fields (repo, homepage, issues, taxonomies)
3. Write content: overview, features, git-annex/DataLad integration, AI readiness assessment
4. Submit a PR

TODOs

• Develop Claude Code SKILL (/conserve.add-tool) for adding new tool entries with proper taxonomies
• Set up GitHub Pages deployment via GitHub Actions
• Create a sample (fully or partially private) deployment at e.g. conserve.centerforopenneuroscience.org
• Integrate Entire.io for ongoing AI session archival during development
• Add comparison matrix page (tool x feature grid)
• Create archival workflow guides (step-by-step for each media type)
• Add con/ceptualization#2 vision page (config-driven archival orchestrator)
• Set up CI to validate content front matter schema
• Create RSS/Atom feed for new tool additions
• Explore MkDocs alternative for deeper DataLad ecosystem alignment
• Add click hyperlinks to Mermaid diagram nodes

Technology

• Static site generator: Hugo
• Theme: Congo (Tailwind CSS)
• Version control: DataLad / git-annex
• Hosting: GitHub Pages (planned)

License

Content is provided under CC BY 4.0.