The Agent Skills Directory

[SAFE]: The skill performs repository maintenance tasks on the flutter/devtools repository using the official GitHub CLI (gh). All scripts and instructions are scoped to this specific, trusted repository.
[SAFE]: The skill implements a robust 'Safety Rule' in the SKILL.md instructions, explicitly directing the agent not to close issues based solely on age or screen updates without verifying the specific bug behavior.
[SAFE]: A critical mandate requires the agent to present all drafted closing comments and issue URLs to the user for explicit approval before any destructive actions (closing issues) are taken.
[COMMAND_EXECUTION]: The skill uses Process.run in Dart and shell scripts to execute gh commands. These are used for read-only operations (searching/viewing) and controlled write operations (closing with comments) following user approval. The logic is consistent with the skill's stated administrative purpose.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitHub issue descriptions and comments. However, it mitigates risk by requiring human review of all generated output before it is posted back to GitHub. The 'Style Constraint' against em dashes is a specific formatting rule, not a security bypass.

closing-obsolete-issues