#Output reference

A normal build produces a combined Markdown library plus companion audit files.

#Default outputs

For an output library called:

markdown-library.md

the companion files are:

Output	When written	Purpose
`markdown-library.md`	Always	Combined AI-readable Markdown library.
`markdown-library-manifest.md`	Always	Human-readable status table for found files.
`markdown-library.index.json`	Default	Machine-readable index, schema `1.2`.
`markdown-library.index.yaml`	With `--index-format yaml` or `both`	Optional YAML version of the index.
`markdown-library-files/`	With `--individual-files`	One generated Markdown file per included source.

Example CLI output:

Done. Markdown library created.
  Library:  /path/to/markdown-library.md
  Manifest: /path/to/markdown-library-manifest.md
  JSON index: /path/to/markdown-library.index.json
  Sources included: 12
  Sources skipped:  2
  Individual files: 12 in /path/to/markdown-library-files

#Combined library

The combined library is deliberately plain Markdown. Each included source is separated by source markers so humans and tools can identify boundaries.

The file also starts with compact YAML front matter. This metadata is for tools such as a future library viewer. It is human-readable in the raw .md file, but many Markdown viewers hide or treat front matter as page metadata.

Example front matter:

---
make_markdown_library:
  type: "library"
  schema_version: "1.0"
  title: "Markdown Library File"
  description: "Client reading pack"
  category: "Client Work"
  full_index: "markdown-library.index.json"
  sources:
    - id: "src_abc123"
      title: "report.pdf"
      relative_path: "report.pdf"
      converter: "liteparse 2.2.1"
      warnings: []
---

The readable source sections still keep the older START/END markers for backward compatibility. Each section also has hidden mmlib:source-start and mmlib:source-end HTML comments so a viewer can find section boundaries without relying only on line numbers.

A typical section contains:

source path
source fingerprint
converter used
converter mode
fallback note, if any
converted or direct-ingested Markdown content

#Manifest

The manifest is for people. It records which files were included, imported, skipped, duplicated, unsupported, or failed.

Use it to answer:

Did this file get included?
Why was that file skipped?
Which converter was used?
Did fallback happen?
Were duplicates detected?

#JSON/YAML index

The index is for automation. Schema 1.2 records:

schema_version
tool name/version
library output path
source input path
build settings
included sources
skipped sources

Each source entry records:

sha256 and short fingerprint;
source path, suffix, size, and parent archive if applicable;
converter, converter_version, converter_mode, and converter_options;
output.char_count, output.line_count, and output.word_count;
fallback.used, fallback.from, fallback.to, and fallback.reason;
complexity.checked, complexity.complex, and complexity.reason;
markdown.policy, markdown.generated, and markdown.library_import;
library_section offsets when the source is included;
individual Markdown output path when split files are enabled.

#Individual files

With:

--individual-files

the tool writes split outputs to the default directory:

markdown-library-files/

Each split file starts with a source marker. Generated split files are skipped on future runs by default.

#Overwrite and backup behaviour

make refuses to replace existing library, manifest, and index outputs unless you choose one of these options:

--backup-existing keeps backup copies before replacing outputs.
--overwrite replaces outputs without backups.

--backup-existing writes backups such as:

markdown-library.backup.md
markdown-library-manifest.backup.md
markdown-library.index.backup.json

add, rebuild, and remove-file back up existing libraries before modifying them by default.

#Individual output collisions

Existing user-authored Markdown is protected. If a generated split file would collide with an existing non-generated Markdown file, the tool chooses a numbered filename unless --overwrite-individual is set.

project.md      # user-authored, preserved
project-2.md    # generated output