The Agent Skills Directory

[SAFE]: The skill serves as a legitimate orchestration tool for project management. Its behavior is consistent with its documentation and intended use-case.
[COMMAND_EXECUTION]: The skill utilizes two local bash scripts (detect-state.sh and recommend-next.sh) to analyze the project environment. These scripts perform read-only operations using standard shell commands (grep, ls, cat) to identify the presence and status of planning documents. The scripts include proper shell variable quoting and directory existence checks to ensure stable execution.
[EXTERNAL_DOWNLOADS]: No external resources, dependencies, or scripts are downloaded from the internet. All logic is contained within the skill's local files.
[DATA_EXFILTRATION]: There are no network-capable tools or commands (such as curl or wget) used within the skill. Data interaction is limited to reading project metadata for the purpose of state inference.
[PROMPT_INJECTION]: The instructional content in SKILL.md is strictly focused on task routing and state detection. It does not contain instructions to bypass AI safety filters or override the agent's core identity.

bmad-help