The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions (Task 3) explicitly direct the agent to search for and read .env* files as part of infrastructure exploration. Since environment files frequently contain hardcoded credentials, API keys, and other secrets, this instruction creates a risk of sensitive data exposure.
[PROMPT_INJECTION]: The skill facilitates the ingestion of large amounts of untrusted data from a codebase into the agent's context through multiple subagents, which presents a surface for indirect prompt injection.
Ingestion points: Files throughout the codebase, including sensitive directories and configuration files specifically targeted by the exploration tasks (e.g., src/, .env*, .github/).
Boundary markers: Absent; no instructions are provided to the agent to delimit untrusted codebase content or to ignore potential instructions embedded within those files.
Capability inventory: Uses the Task tool with subagent_type: "explore", which is a read-only code search and analysis capability.
Sanitization: Absent; the skill does not include steps to sanitize or validate the content retrieved from the codebase before synthesizing the final report.

parallel-exploring