Browse / LLM wiki / wiki-knowledge
wiki-knowledge
dhague
Git-enhanced LLM-Wiki optimised for LLM efficiency
Open dhague/wiki-knowledge → On the boundary of the four rules.
Works with any agent
ships AGENTS.md, the cross-tool instruction file.
In its own words
Personal knowledge management powered by LLM agents. Turns raw documents into a structured, searchable, git-backed markdown wiki vault — then answers questions over it with typed-edge graph traversal and cited synthesis.
Follows the Karpathy LLM-wiki pattern.
What's inside
- wiki-knowledge — a host-neutral Agent Skills package (Claude Code, OpenCode, DeepSeek Harness, and any host that reads the standard's skill directory) providing ingestion and retrieval over a markdown wiki vault
- Agent pipeline — on Claude Code, Sonnet for semantic ingestion (chunking, overlap classification, edge typing) and Haiku for retrieval (query expansion, BM25 search, frontier traversal, synthesis); every other host runs the same procedures on its session model
- Deterministic script layer — a single TypeScript bundle for vault I/O, placement, FTS5 search indexing, and commit construction (no model calls, no runtime to install — it runs on the Node or Bun the host already has)
- Full-text search — SQLite FTS5 via stdlib, zero extra search dependencies
Install
Any host (Agent Skills package)
Installs all eight skills into the host's own skill directory — .agents/skills/ for OpenCode and DeepSeek Harness (which reads /.agents/skills natively), .claude/skills/ for Claude Code, and the equivalent for the other hosts the skills CLI supports. Add -g to install for your user rather than the current project.
Quoting 16% of the README. Read the rest on GitHub ↗
Read from the repository's git tree on 2026-08-17.