The Agent Skills Directory

[COMMAND_EXECUTION]: Executes standard shell commands (ls) for project type detection and git commands (git log, git diff) to analyze recent changes. These are typical operations for an automated debugging tool.
[PROMPT_INJECTION]: Ingests external data through $ARGUMENTS (error messages) and project files like AGENTS.md. This creates an indirect prompt injection surface where malicious content in a bug report or project file could attempt to influence the agent's logic.
Ingestion points: User-provided $ARGUMENTS in SKILL.md, project-level AGENTS.md, and git log history.
Boundary markers: No explicit delimiters or instructions are used to isolate untrusted data.
Capability inventory: The skill can execute git and ls commands, read files within the lessons directory, and invoke other specialized sub-agents.
Sanitization: No sanitization is performed on error messages or file contents before they are processed by the LLM.
[EXTERNAL_DOWNLOADS]: References a 'web-research agent' for solution searching. This utilizes the platform's internal agent capabilities rather than performing direct network requests to untrusted sources.

debug