Skip to content

Repository files navigation

Degardis

Degardis is a command-line compiler for authors of portable Agent Skills. It turns structured YAML sources into self-contained bundles for Claude, Codex, Copilot, Cursor, Roo, and ChatGPT. Before writing an artifact, it validates the source, selected profiles, generated paths, and generated links.

Authoring stays separate from generated output: edit the source files, validate them, and rebuild either an installable folder or a distributable ZIP.

Quick start

Degardis requires Python 3.10 or later. Install it from PyPI:

python -m pip install degardis
degardis --version

For an isolated command-line installation, use pipx install degardis.

To work on Degardis from this repository instead:

python -m pip install -e .
degardis validate examples/structured-summary
degardis list examples/structured-summary
degardis build examples/structured-summary --output .artifacts

The successful build ends with Summary: 1 skill built as folder. and reports the artifact path. The generated folder is ready to inspect or install:

.artifacts/structured-summary/
  SKILL.md
  agents/openai.yaml
  references/
  scripts/
  assets/

Degardis writes only beneath the requested output root. Rebuilding a skill replaces that skill's existing folder and ZIP there, while preserving unrelated entries. See Artifact format before building directly into an agent's skill directory.

Add the example's optional profile or produce a ZIP:

degardis build examples/structured-summary --profile detailed --output .artifacts
degardis build examples/structured-summary --zip --output .artifacts

examples/structured-summary is the canonical public example used throughout the guides and CLI help. It summarizes supplied material from any subject; synthetic skills under tests/fixtures are compiler test data.

Companion skills

The Degardis skills catalog contains reusable skills authored with Degardis.

For the smoothest authoring experience, use Degardis Authoring. It guides a skill from initial design through review, validation, packaging, and installation while applying Degardis conventions across manifests, workflows, entries, profiles, scripts, and assets.

Commands

degardis list PATH [PATH ...]
degardis validate PATH [PATH ...]
degardis build PATH [PATH ...] --output PATH [--profile [SKILL:]PROFILE] [--zip]

Run degardis COMMAND --help for exact options and examples. All commands accept individual skill directories, directories containing skills at any depth, or a mixture of both. This lets a source tree group skills into subdirectories while building all of them with one command.

An unqualified profile name applies to every selected skill that defines it. SKILL:PROFILE selects one owner, and all selects all available profiles. Supplying explicit selectors replaces manifest defaults for that build.

For exact command behavior, profile selectors, source schemas, and the Python API, see the reference.

Documentation

Reader goal Document
Build and install the example Getting started
Understand the source and artifact models Concepts
Create or modify a skill Authoring guide
Look up commands, schemas, or the Python API Reference
Inspect and install generated output Artifact format

Development

python -m unittest discover -s tests -v
python -m degardis validate examples/structured-summary
python -m degardis build examples/structured-summary --profile all --output .artifacts

About

Compile validated Agent Skill sources into portable, installable bundles for Claude, Codex, Copilot, Cursor, Roo, and ChatGPT.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Contributors

Languages