The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests and processes user-supplied $ARGUMENTS to determine its operational mode and capture issue descriptions.
Ingestion points: The $ARGUMENTS variable in SKILL.md is the primary entry point for untrusted user input.
Boundary markers: Absent; the skill interpolates user data directly into the execution context without delimiters or explicit instructions for the agent to ignore embedded commands.
Capability inventory: The skill has access to powerful tools including Bash, Read, Write, and Agent, and it includes logic to execute a bash command (ls .planning/debug/*.md) for session management.
Sanitization: No input validation, filtering, or sanitization of the $ARGUMENTS string is performed.
[SAFE]: The skill references a local configuration file path (~/.claude/gsd-core/workflows/debug.md) that is consistent with the vendor's own application structure and intended debugging purpose.

gsd-debug