The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several shell commands to automate the scaffolding process. These include directory management via mkdir -p, version control operations via git mv and git commit, and the execution of a project-specific linter using pnpm ai-hero-cli internal lint.
[PROMPT_INJECTION]: The workflow relies on parsing an external "plan" to determine section and exercise names, which introduces a surface for indirect prompt injection. If the plan contains untrusted or malicious input, it could potentially influence agent behavior or attempt command injection if the agent does not properly escape the extracted strings.
Ingestion points: The agent extracts section names, exercise names, and variant types from a user-provided plan.
Boundary markers: The skill does not define specific boundary markers or instructions to treat the plan content as untrusted data.
Capability inventory: The skill leverages shell command execution (mkdir, git, pnpm) and file system write operations.
Sanitization: There are no instructions for validating or sanitizing the input from the plan before it is used in shell commands or file creation.

scaffold-exercises