fleet-sync
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill clones a remote repository from
https://github.com/filipgutica/filip-stack.gitand executes JavaScript files (e.g.,scripts/sync.mjs,field-guide.mjs) using thenoderuntime. These scripts are responsible for the core synchronization logic and represent vendor-owned resources from the skill author. - [COMMAND_EXECUTION]: The skill invokes several external command-line tools to perform its tasks, including
gitfor repository management,nodefor script execution,chezmoifor dotfile preference transport, and thefilip-stacktool itself. - [INDIRECT_PROMPT_INJECTION]: The skill processes external configuration data from
fleet.jsonand preference files likefield-guide.json. Maliciously crafted content in these files could theoretically influence the agent's behavior during the synchronization process. - Ingestion points: Reads policy data from
~/.config/filip-stack/fleet.jsonand preference data viachezmoifromfield-guide.json. - Boundary markers: The instructions specify the use of schema validation contracts such as
validatePortableSharedGuidanceandschemas/annotations-v1.schema.jsonto verify input integrity. - Capability inventory: The skill has access to shell command execution (
node,git,chezmoi), file system writes for configuration backups, and cross-environment task delegation. - Sanitization: Verification of repository identity, branch cleanliness, and schema-based validation of portable formats are mandated before processing data.
- [EXTERNAL_DOWNLOADS]: The skill performs external network operations to clone the
filip-stackrepository from GitHub and interacts with a plugin marketplace located atfilipgutica/design-system.
Audit Metadata