gitsite-content-manager
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a local Python script
scripts/manage_content.pydesigned to automate file system operations like directory creation and content deletion. - [INDIRECT_PROMPT_INJECTION]: The script ingests user-provided text (e.g., titles, descriptions) and interpolates it into Markdown files and the
site.ymlconfiguration. - Ingestion points: CLI arguments passed to
manage_content.py(e.g.,--title,--description). - Boundary markers: Markdown headers and YAML structures are used to encapsulate input in the generated files.
- Capability inventory: The script uses
pathlibfor file/directory creation andshutil.rmtreefor content deletion within thescripts/manage_content.pyfile. - Sanitization: While a
_slugifyfunction is used for directory names, input interpolated into thesite.ymlnavigation list is not fully escaped, which could lead to malformed YAML if malicious characters are injected. - [METADATA_POISONING]: The
SKILL.mddocumentation claims the tool generates complete YAML frontmatter, but the implementation inscripts/manage_content.pyexplicitly omits it and states that the platform does not support it. This inconsistency between the skill's stated capabilities and its actual behavior could mislead users or agents.
Audit Metadata