Texuli reads untrusted Markdown and HTML into one AST, writes a safe HTML fragment, and converts chat dialects without raising trust. The stack lives in @texuli/core; these subpaths were released in 0.18.0.
Status
In development (pre-1.0), released in 0.18.0. Agents should call md.convert and md.render rather than import the texels, unless they are writing app code.
The overview for app authors is the core Markdown README.
Subpaths
@texuli/core/markdownParse GitHub Flavored Markdown or strict CommonMark 0.31.2 into the texuli-md/v1 AST, and write it back.Package docs@texuli/core/markdown-htmlWrite the AST as a safe HTML fragment through core/markup. Raw HTML is dropped; every href and src uses the URL policy.Package docs@texuli/core/htmlRead untrusted HTML into the AST with a DOM-free tokenizer and an element and attribute allowlist.Package docs@texuli/core/markdown-dialectsConvert the AST to and from Slack, Discord, WhatsApp, Telegram, Teams, Google Chat, Mattermost and plain text, with loss diagnostics.Package docs
Agents
md.convert and md.render are edge operations on the local MCP (npx -y @texuli/mcp), the remote Texuli MCP, and texuli md convert / texuli md render. Related Solutions: Agent chat and Rich text editor.