framework 1.1.0
plumber moved from Imports to Suggests. plumber pulls
in sodium, which needs the system libsodium library and
often fails to install on HPC clusters. The GUI (gui(),
setup()) now asks you to install plumber when you launch
it; nothing else needs it
- New standard
logs directory for run logs and cluster
job output. All project gitignore templates now ignore
**/logs/** and slurm-*.out, and
security_audit() treats logs as a private
directory
scaffold() git checks are now portable across
platforms
- Skills-based AI context: new projects get a thin canonical
AGENTS.md plus Framework skills installed in
.claude/skills/ (workflow, data, packages, outputs, and
sensitive-data rules as separate SKILL.md files)
AGENTS.md replaces CLAUDE.md as the
default canonical AI context file; CLAUDE.md and
.github/copilot-instructions.md are now thin pointer stubs
- New
ai_skills_update() refreshes a project’s Framework
skills after upgrading the package
ai_regenerate_context() falls back to
CLAUDE.md for projects created before 1.1.0
framework 1.0.0
- First stable release
- Zero CRAN check errors/warnings (1 acceptable note for global config
object)
- Complete documentation for all exported functions
- Comprehensive test suite (300+ tests)
framework 0.10.1
- Fixed non-ASCII characters in R source files for CRAN
compatibility
- Fixed documentation for
subdir parameter in
make_rmd(), make_revealjs(), and
make_presentation()
- Added missing package declarations to DESCRIPTION Suggests
- Improved
scratch_capture() examples with
\dontrun{}
framework 0.10.0
- Initial preparation for CRAN submission
- Comprehensive data management system with declarative YAML
catalogs
- SQLite, PostgreSQL, MySQL, DuckDB database connectivity
- Quarto-first notebook generation with stub templates
- S3-compatible object storage publishing
- Project scaffolding with reproducible environments
- Caching system with expiration support
- Git hooks integration for data security
- GUI for project management (via
gui())