markymark
A language server and AI agent tool for Markdown and structured data files.
Rust workspace with a Zig SIMD parser core. Provides navigation, refactoring,
search, and diagnostics across your workspace — as an editor extension (LSP)
and AI agent tool (MCP).
Pre-release: APIs and behavior may change before v1.0.
Features
Navigation
- Go to definition — jump to any heading, wiki link, or block ID across files
- Find all references — see every document linking to a heading or symbol
- Rename everywhere — rename headings and update all references in one step
Search
- Symbol search — fuzzy-match headings, tags, and code references (Ctrl+T / Cmd+T)
- Semantic search — rank sections by relevance using embeddings (Voyage API or local ONNX)
- Full-text search — free text with frontmatter, tag, and property filters
- Regex patterns — search file content with context lines and glob filters
Diagnostics
- Broken link detection — wiki links, Markdown links, and heading anchors validated as you type
- Link graph analysis — orphan documents, hub detection, and connected cluster mapping
Formats
- Markdown, JSON, JSONC, JSON5, JSONL, YAML, TOML, .env, INI
- Obsidian + Logseq — wiki links, callouts, block IDs, and page properties
Under the Hood
markymark is a seven-crate Rust workspace with a Zig FFI layer:
- Zig parser core — dependency-free md4c implementation with SIMD-accelerated extraction, statically linked into the final binary
- Cross-document index — string-interned lookup tables with O(1) wiki link resolution, incremental contribution-diffing, and lazy tag maintenance
- Dual protocol — LSP for real-time editing and MCP for AI agent workflows, backed by the same index
- Embedding search — optional semantic index using Voyage API or local ONNX models via fastembed
Install
cargo install markymark-cli
Pre-built binaries: GitHub Releases
Documentation
Full documentation at sethyanow.github.io/markymark:
Claude Code
Install the plugin for LSP + MCP document intelligence:
claude /plugin install markymark
Add to your CLAUDE.md for LSP-first document reading:
## Document Intelligence
This project uses markymark LSP. Prefer LSP over reading raw files:
- `LSP documentSymbol <file>` — structure/outline before Read
- `LSP hover <file> <line> <col>` — heading backlinks, key path info
- Diagnostics (broken links, duplicate headings) reported automatically
- Works for Markdown, JSON, YAML, TOML, .env, INI, and more
License
AGPL-3.0
sethyanow