Manifest template for FAIR data package
Use this template to create a structured inventory of all files included in a FAIR data package. The manifest records file names, locations, formats, roles, relationships, access conditions and integrity information, helping users understand and verify the package.
Resource information
Resource type
File inventory template
Intended users
Researchers, data curators, data stewards and repository managers
Recommended use
FAIR data packaging, repository deposit and dataset preservation
Status
Recommended template of the Competence Center
Download the manifest template
Choose the format that best matches your workflow. CSV is recommended as the basic interoperable format. XLSX is convenient for manual preparation and review, while JSON is suitable for machine-readable data packages and automated workflows.
Available formats
What is a manifest?
A manifest is a structured inventory of the files included in a data package. It records what each file is, where it is located, which role it performs, how it relates to other files and, where applicable, how its integrity can be verified.
The manifest complements the README. The README explains the dataset and provides instructions for users, while the manifest gives a systematic, file-level description of the package.
README and manifest have different functions
The README explains the dataset and provides instructions for users. The manifest gives a systematic, file-level description of the package.
What the template covers
The template organises file-level information into four practical documentation areas.
File identification
File name, relative path, file extension, media type, file size and package location.
File role and content
Short description, file category, data processing level, responsible software and the function of the file within the package.
Relationships and provenance
Source file, derived file, related workflow step, dependency, input-output relationship and associated documentation.
Integrity and access
Checksum, checksum algorithm, access category, licence, sensitivity status and additional restrictions.
Recommended manifest fields
The exact set of fields may be adapted to the dataset, discipline and repository requirements.
| Field | Purpose | Example |
|---|---|---|
file_name |
Name of the file including its extension | simulation_results.csv |
relative_path |
Location of the file within the package | results/simulation_results.csv |
file_role |
Function of the file within the package | processed data |
description |
Short explanation of the file content | Calculated values used in the final analysis |
format |
File format or media type | text/csv |
size_bytes |
File size in bytes | 245760 |
related_to |
Related input, output or documentation file | scripts/analyse.py |
checksum |
Value used to verify file integrity | 8f14e45f... |
checksum_algorithm |
Algorithm used to generate the checksum | SHA-256 |
access |
Access category or restriction | open |
licence |
Licence applicable to the file | CC BY 4.0 |
How to use the template
1. Prepare
Organise the dataset and identify all data, scripts, documentation and result files.
2. List
Add one manifest record for every file included in the package.
3. Describe
Specify the content and role of each file.
4. Relate
Record relevant input-output, dependency and derivation relationships.
5. Verify
Check paths, formats, checksums, licences and access conditions.
Example package structure
The manifest should normally be placed in the root directory of the data package.
dataset-name/ │ ├── README.md ├── manifest.csv ├── metadata.json │ ├── data/ │ ├── raw/ │ └── processed/ │ ├── scripts/ ├── configuration/ ├── documentation/ └── results/
Example manifest records
file_name,relative_path,file_role,description,format,related_to,access input.csv,data/raw/input.csv,raw data,Original input observations,text/csv,,open clean.py,scripts/clean.py,processing script,Script used to clean the input data,text/x-python,data/raw/input.csv,open cleaned.csv,data/processed/cleaned.csv,processed data,Cleaned observations used for analysis,text/csv,scripts/clean.py,open README.md,README.md,documentation,Documentation for the complete dataset,text/markdown,,open
Important notes
-
Use relative paths rather than local computer paths such as
C:\Users\.... - Use one record for each file, not one record for each folder.
- Keep file names and paths exactly consistent with the actual package.
- Describe file roles consistently, for example: raw data, processed data, script, configuration, metadata, documentation and result.
- Do not include passwords, access credentials or confidential information in the manifest.
- Update the manifest whenever files are added, removed, renamed or replaced.
- Generate checksums only after the final versions of the files have been prepared.
Minimum manifest
At minimum, the manifest should contain one record for each file and identify its name, relative path, role, short description and format.
Recommended minimum fields:
file_name,
relative_path,
file_role,
description and
format.