AI Skill Reference

Markmeld melds markdown/YAML via Jinja2 templates → pandoc → documents. Philosophy: content + structure + style → output.

Install & CLI

pip install markmeld   # requires pandoc
mm                     # list targets
mm <target>            # build target
mm <target> -p         # preview (no build)
mm <target> -d         # dump template data
mm <target> -e         # explain target

Configuration (_markmeld.yaml)

targets:
  manuscript:
    jinja_template: template.jinja
    output_file: output/{today}_manuscript.pdf
    command: pandoc -o "{output_file}"
    data:
      md_files:
        content: src/manuscript.md
      yaml_files:
        metadata: src/metadata.yaml
      variables:
        author: "Jane Doe"

Data sources: md_files, yaml_files, variables, md_globs, yaml_globs

Target types: standard (template→command), type: raw (command only), type: meta (orchestrates others)

Inheritance: inherit_from: base_target

Side targets: prebuild: [...], postbuild: [...]

Jinja Templates

---
title: {{ title }}
---
{{ content }}

Special variables: {{ _today }}, {{ _now }}, {{ target_name }}, {{ _md[key].content }}, {{ _yaml[key].content }}

Multi-output (mail merge)

loop:
  loop_data: recipients   # list from YAML
  assign_to: recipient    # available in template
output_file: "letter_{recipient.name}.pdf"

Advanced Features

• Imports: imports: [/path/to/_markmeld.yaml]
• Google Drive: type: google-doc with google_docs: {doc_id: "..."} (needs MM_GOOGLE_DRIVE_CREDENTIALS)
• Remote templates: jinja_template: https://databio.org/mm_templates/v1/...
• Pandoc filters: figczar (figures), change-marker (track changes), multi-refs (bibliographies)

Docs

Full documentation: https://sciquill.databio.org/markmeld/<page>.md (raw markdown, AI-fetchable)