The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard local development commands such as pnpm test, pnpm lint, pnpm typecheck, and various git subcommands (add, commit, checkout, stash). These operations are consistent with the skill's stated purpose of automating the TDD cycle within a local development environment.
[DATA_EXPOSURE]: The skill is designed to read project-specific files, including design documentation (docs/design-docs/), source code (packages/**/src/**/*.ts), and existing test files. This data access is localized to the workspace and is required for the agent to understand requirements and implement features.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes content from design documents and source files which may be external or untrusted.
Ingestion points: Reads content from docs/design-docs/**/*.md, packages/**/src/**/*.ts, and **/*.test.ts via Glob and Read operations.
Boundary markers: The skill uses structured labels such as 【Design Docs（関連部分）】 and 【収集した要件】 in its prompt templates to separate external data from system instructions.
Capability inventory: The skill can execute shell commands through the package manager (pnpm), perform git commits, and modify project files.
Sanitization: No specific content sanitization or validation of external text files is explicitly mentioned, relying on platform-level guardrails.

tdd-from-design-docs