The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various local system commands and project-specific scripts to manage the update process.\n
Uses subprocess.run to invoke Git for repository state inspection (git status, git diff, git ls-remote).\n
Executes bd (Beads CLI) for project health checks and hook management.\n
Runs project-local scripts scripts/setup-tooling.py and scripts/check-docs.py following template updates.\n- [EXTERNAL_DOWNLOADS]: The skill fetches template data from an external source.\n
Downloads project templates from the author's GitHub repository (gh:RobertDeRose/dstack) using the Copier library.\n
The implementation explicitly sets unsafe=False when calling copier.run_update and copier.run_copy, which prevents the execution of arbitrary shell tasks defined within the template metadata.\n- [DYNAMIC_EXECUTION]: The skill dynamically loads Python logic from the project environment.\n
Uses importlib.util to load layout_contract.py and beads_hooks.py from specific paths within the skill and project structure.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data (e.g., .copier-answers.yml, task files) and remote template content.\n
Ingestion points: Reads project configuration files and external template repositories.\n
Boundary markers: The workflow requires a manual 'path-accounting ledger' as described in SKILL.md where every change must be classified (accepted, preserved, or resolved) before committing.\n
Capability inventory: Includes file system writes, Git operations, and execution of setup scripts.\n
Sanitization: Implements strict path validation in layout_contract.py, ensuring paths are relative, normalized, and do not resolve through symlinks.

update-project