The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses bun install --frozen-lockfile within the scripts/bootstrap.ts file to download and install Node.js packages at runtime, which are subsequently imported and executed by the skill's tools.
[COMMAND_EXECUTION]: The skill extensively uses subprocess calls to execute shell commands and external CLI tools including git, gh, du, jq, and trash across multiple playbooks and scripts to perform file system and repository operations.
[DYNAMIC_EXECUTION]: The scripts/bootstrap.ts script re-executes the current process using Bun.spawnSync after ensuring dependencies are installed, creating a dynamic execution flow.
[EXTERNAL_DOWNLOADS]: The skill downloads third-party packages from the npm registry via the bun install command triggered during initialization.
[INDIRECT_PROMPT_INJECTION]: The skill's playbooks (e.g., playbooks/babysit.md) process untrusted data from external sources such as pull request comments and session transcripts. Although the skill includes instructions to treat this data as untrusted, the ingestion of external content into the agent's context creates a potential attack surface.

poteto-mode