The Agent Skills Directory

[SAFE]: The skill follows a legitimate and transparent workflow consisting of context preparation, user interview, and output generation.
[SAFE]: Exploration patterns defined in references/exploration-patterns.md utilize common, non-destructive search tools (grep, glob, find) to analyze appropriate project files such as documentation and configuration files without accessing sensitive system paths or credentials.
[SAFE]: External references in the instructions point to well-known and trusted technology documentation sources.
[PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection because it ingests data from local project files (markdown and configuration files) to prepare for the interview. This content could potentially contain instructions intended to influence the agent; however, the skill's output is intended for human review, which provides necessary oversight.
Ingestion points: Codebase exploration targets such as **/*.md and **/config.* used in Phase 1 (Preparation).
Boundary markers: The skill does not explicitly define delimiters to separate read file content from its own instructional context.
Capability inventory: Performs codebase searches (grep, rg, glob), file reading, and structured user interaction via AskUserQuestion.
Sanitization: No specific sanitization or filtering logic is applied to the content of the ingested files.

interview