The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it is designed to read and process existing project files (such as manifests, source code, and lockfiles) to inform the scaffolding and adaptation process.
Ingestion points: The agent reads existing project instructions, manifests, wrappers, and code files (SKILL.md, references/verification.md).
Boundary markers: Decisions are documented in a structured scaffold-spec.md file to maintain a clean record of project state.
Capability inventory: The skill uses Bash for environment inspection and verification, Write and Edit for creating project files, and the Agent tool for sub-tasks.
Sanitization: The skill includes explicit security guidelines to avoid interpolating untrusted tool input into shell code and specifically instructs the agent to skip reading secret values (references/verification.md).
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform legitimate environment checks, execute build tools (e.g., Gradle, Maven), and run verification tests. These operations are restricted to the local project directory and follow documented project workflows.

scaffold