The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the $ARGUMENTS parameter to execute shell commands such as gh issue view. The absence of input validation or sanitization for the issue identifier creates a risk of command injection, where an attacker could provide a malicious string to execute arbitrary code in the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data which could be used to manipulate agent behavior in a multi-step execution chain.
Ingestion points: Reads external data from GitHub issues and project-specific markdown/HTML files (e.g., research.md).
Boundary markers: The skill lacks explicit delimiters or instructions to prevent the agent from following directives embedded within the ingested data.
Capability inventory: Uses tools like Bash, Write, and Edit to generate implementation plans (plan.md) that direct the actions of subsequent automation skills.
Sanitization: External content is processed and incorporated into the implementation plan without sanitization or filtering.
[EXTERNAL_DOWNLOADS]: Fetches issue data and acceptance criteria from GitHub's official service via the command line.

plan