@markdownai/renderer

Format modules for MarkdownAI directive output. Takes a string[] and produces formatted markdown.

Root README · Engine · GitHub

Install

npm install @markdownai/renderer

Formats

12 format types. The renderer is consumed internally by @markdownai/engine and the MCP server; you'd reach for it directly only when building tooling that formats data outside a full render pass.

| type | Output | |---|---| | list | - item bullets | | numbered | 1. 2. ordered items | | links | - [label](url) per item | | table | Markdown grid table | | code | Fenced code block | | inline | Plain scalar, no wrapping | | bar | ASCII horizontal bar chart | | flow | a -> b -> c flow diagram | | tree | ASCII indented tree | | timeline | Left-to-right ASCII timeline | | json | Pretty-printed JSON in a fenced block | | row | (new in v2) JSON block for a single document |

The row format

row formats a single object as a JSON code block. Pair it with @db ... as=row label=X visible=false to capture a struct into ctx.data without rendering anything inline:

@db
  using="local"
  find="features"
  where='id == "auth"'
  as=row
  label=feature
  visible=false
@db-end

{{ feature.title }} has {{ feature.source_files.length }} files

For full @db / @read usage with struct labels, see the engine README.

API

import { render, aiFilter } from '@markdownai/renderer'
import type { RenderType, RendererInput, AiFilterOptions } from '@markdownai/renderer'

render({ type: 'table', data: ['name | status', 'parser | ok'], columns: ['name', 'status'] })
render({ type: 'row', data: ['{"id":"auth","title":"Authentication"}'] })

const compressed = aiFilter(rendered)  // post-render pass that drops decorative whitespace

RendererInput is { type: RenderType, data: string[], columns?: string[], options?: Record<string, string> }.

License

MIT.