The Agent Skills Directory

[PROMPT_INJECTION]: The skill directly interpolates user-supplied $ARGUMENTS into the TARGET section of the prompt. This creates a surface for direct prompt injection where a user could provide input designed to override the agent's core instructions or redirect its behavior.- [DATA_EXFILTRATION]: The skill is designed to access and write to the ~/.claude/projects/ directory for telemetry logging. Interacting with hidden, application-specific directories in the user's home folder is a sensitive file path access pattern that involves platform-level metadata.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it analyzes external codebase content (Step 1.2, 2.1) and processes it as part of the agent context without implementing boundary markers to separate data from instructions. Ingestion points: Local codebase files, package manifests, and configuration files. Boundary markers: Absent; no specific delimiters or ignore-instructions warnings are used when processing external files. Capability inventory: The skill can write files to the docs/ directory and append telemetry to ~/.claude/. Sanitization: Absent; the skill does not show evidence of sanitizing or escaping content retrieved from the analyzed codebase.

teacher-workload