The Agent Skills Directory

[SAFE]: The skill's primary function is to provide a planning and grounding framework for software development tasks. It uses instructions to guide the agent toward safe, read-only investigation and evidence citation.
[SAFE]: The SKILL.md file contains instructions that explicitly mandate 'read-only investigation' and 'safe non-mutating checks,' effectively mitigating risks associated with unauthorized code modification or command execution.
[SAFE]: The skill uses a .workflow/ directory for sidecar documentation, which is a localized, non-executable, and standard practice for recording task state within a development environment.
[SAFE]: While the skill instructions include some imperative language (e.g., 'IMPORTANT', 'CRITICAL'), these are used in the context of methodology compliance rather than attempting to bypass the agent's safety guardrails or security filters.

task-groundwork