The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: During initial setup, the skill installs the Bun runtime by downloading a shell script from the official bun.sh domain and piping it directly to bash. This follows a standard but high-privilege installation pattern for developer tools.
[DATA_EXFILTRATION]: The skill implements a telemetry system to collect data such as skill usage, session duration, and success outcomes. It utilizes the vendor-provided gstack-telemetry-log utility for transmission. The skill proactively asks the user for permission and offers community or anonymous modes before enabling this feature.
[COMMAND_EXECUTION]: The skill executes various local bash scripts and binaries related to the gstack toolsuite, primarily for session management, configuration updates, and telemetry logging. These resources are located in the user's home directory.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and analyzes content from external, user-provided URLs to determine necessary source code modifications. * Ingestion points: HTML, CSS, and JavaScript content retrieved from target websites via the browse tool during audit phases. * Boundary markers: The skill does not implement explicit delimiters or instructions to the agent to disregard embedded commands within the external web content. * Capability inventory: The agent possesses significant capabilities, including shell execution, file system modification, and git repository management. * Sanitization: There is no explicit sanitization of the content retrieved from websites before it is used as context for the agent's reasoning and subsequent code edits.
[EXTERNAL_DOWNLOADS]: The skill fetches the Bun runtime installer from bun.sh and references external essays and documentation from garryslist.org during the onboarding process.

design-review