The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute a local metrics script (./scripts/analyze-codebase.sh) during the complexity breakdown phase. This is an expected behavior for a code assessment tool.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the way it processes external files from the codebase.
Ingestion points: Data enters the agent context through the Read, Grep, and Glob tools when evaluating the target specified in arguments.
Boundary markers: Sub-agent prompts include a ## Scope Constraint section to limit file access, but they do not use specific delimiters or explicit 'ignore instructions' warnings to prevent the LLM from obeying commands embedded within the source code being analyzed.
Capability inventory: The skill possesses significant capabilities, including Bash execution, TaskCreate for spawning sub-agents, and file system access via Read and Grep.
Sanitization: There is no evidence of content sanitization or filtering of the ingested file data before it is passed to the analysis agents.

assess