Heinrich (@arscontexta) — Knowledge Systems & Context
Posted: January 18, 2026 Engagement: 84 replies · 265 reposts · 2.6K likes · 8.7K views
The Article
Heinrich published an X Article (long-form post) about AI knowledge systems and structured context. The article is behind X’s paywall, but follow-up posts provide context:
“everything is a context problem — when people say AI can’t do real work, what they’re actually saying is they gave it bad context”
Key themes from his thread:
- Structured context graphs that agents can traverse
- Knowledge systems derived from conversation
- Skill graphs > traditional SKILL.md files
arscontexta — Claude Code Plugin
The article’s subject is likely about arscontexta, Heinrich’s Claude Code plugin:
GitHub: agenticnotetaking/arscontexta — 3.4K ★
Claude Code plugin that generates individualized knowledge systems from conversation. You describe how you think and work, have a conversation and get a complete second brain as markdown files you own.
How it works
- Conversation → Knowledge system — describe how you think and work
- Plain markdown vault — wiki-linked files forming a traversable knowledge graph
- No database, no cloud, no lock-in — you own the files
- Processing pipeline — skills that extract insights, find connections, update old notes, verify quality
Plugin Architecture
arscontexta/
├── .claude-plugin/
├── skills/ # 10 plugin-level commands
│ ├── setup/ # Conversational onboarding
│ ├── ask/ # Query the research graph
│ ├── health/ # Diagnostic checks
│ ├── reduce/ # Extract insights
│ ├── reflect/ # Find connections
│ └── ...
└── skill-sources/ # Generated command templates
Related
- LLMs from Scratch
- Understand Anything — code→knowledge graph
- AI Memory Part 2 — MLDoE