skill-forge
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a suite of Python scripts (
init_skill.py,validate_skill.py,package_skill.py,repo_lint.py,sync_manifests.py,sync_readme.py) designed to manage the local development lifecycle of agent skills. - The
SKILL.mdfile andreferences/refactoring.mdcontain explicit 'execution gates' instructing the agent to only run these scripts when the user provides direct authorization. tests/run-validation.pyusessubprocess.runto execute the local packaging script during automated tests, which is a standard development practice.- [DYNAMIC_EXECUTION]: The test suite in
tests/run-validation.pyutilizesimportlib.utilto dynamically load the skill's ownvalidate_skill.pymodule for unit testing. - [OBFUSCATION]: The
scripts/validate_skill.pymodule proactively defends against homoglyph attacks by usingunicodedata.normalize('NFKC', name)to normalize skill names during validation. - [DYNAMIC_EXECUTION]: The
scripts/validate_skill.pyfile includes a custom_minimal_yamlparser designed to safely handle a subset of YAML frontmatter when thePyYAMLlibrary is unavailable, avoiding dangerous deserialization patterns.
Audit Metadata