The Agent Skills Directory

[SAFE]: The skill follows secure patterns for a development tool. It separates planning from implementation, requires user approval for generated specifications, and limits its persistence to the project's .specs/ directory.
[PROMPT_INJECTION]: The researcher agent (agents/researcher.md) ingests codebase content and web data, which introduces a surface for indirect prompt injection. This risk is inherent to the tool's core functionality of codebase analysis and is mitigated by a human-in-the-loop workflow where the user reviews research results and interview questions before implementation begins.
[COMMAND_EXECUTION]: The skill uses the Bash tool to identify project structure and run developer-defined test suites. These actions are standard for development agents and are restricted to identifying project topology or verifying code changes as explicitly defined in the spec.
[EXTERNAL_DOWNLOADS]: The skill documentation references standard installation procedures using public package managers (npx) and official repositories associated with the author (ngvoicu).

specmint-core