The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill relies on reading external specification files (.specs/SPEC-*.md) to guide its logic, which presents a surface for indirect instructions to be processed by the agent.
Ingestion points: Requirements and acceptance criteria are extracted from local markdown files located in the .specs/ directory (referenced in SKILL.md).
Boundary markers: The skill lacks explicit instructions to ignore or isolate prompt-like content within the specification files; instead, it defines the SPEC as the "single source of truth."
Capability inventory: The agent has the capability to write and modify project files and execute shell commands to run test suites and linters.
Sanitization: There are no mechanisms described for sanitizing or validating the contents of the specification files before they are interpreted.
[COMMAND_EXECUTION]: The skill instructions require the agent to execute shell commands for running test suites, linting tools, and type checkers (referenced in Step 6 of SKILL.md). These are standard operations for a development-focused skill but involve executing tools against code that the agent has just generated.
[DYNAMIC_EXECUTION]: The skill follows a red-green-refactor workflow where it generates test and implementation code and then immediately executes that code using a test runner. This is the intended behavior for TDD automation.

tdd-spec