The Agent Skills Directory

[SAFE]: The skill consists entirely of Markdown documentation and configuration files that describe the "red-green-refactor" loop, seam identification, and mocking strategies. No executable code is bundled with the skill.
[INDIRECT_PROMPT_INJECTION]: The skill instructions suggest that the agent read GLOSSARY.md and Architecture Decision Records (ADRs) to ensure terminology matches the project's domain.
Ingestion points: Reads GLOSSARY.md and ADR files (referenced in SKILL.md).
Boundary markers: Absent.
Capability inventory: Mentions calling the codebase-design skill; the skill itself does not perform file writes, network operations, or command execution.
Sanitization: Absent.
[EXTERNAL_DOWNLOADS]: The UPSTREAM.md file contains instructions for installing the skill from the author's repository (light-cheng/skills-3rdParty). This is part of the standard installation and update process for the skill collection.

tdd