The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes external PowerShell scripts (update-wrapper.ps1 and update-orchestrator.ps1) using pwsh. These scripts are prerequisites for the skill's functionality but are not provided in the analyzed material, which means the agent executes unverified logic from the local filesystem.
[EXTERNAL_DOWNLOADS]: The skill fetches update templates and configuration from GitHub. While GitHub is a well-known service, the skill downloads this remote content and applies it directly to the filesystem, including sensitive directories like .claude/.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection through external data ingestion. • Ingestion points: Data is retrieved from the GitHub API and local .specify/manifest.json files. • Boundary markers: No explicit markers or instructions to ignore embedded commands are present in the documentation. • Capability inventory: The skill has the ability to write to the filesystem and execute shell commands via PowerShell. • Sanitization: The workflow uses file hashing for integrity but does not specify sanitization for remote data retrieved from the API before it is processed.

speckit-updater