setup-jd-solanki-skills
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEPERSISTENCEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PERSISTENCE]: The skill instructs the agent to modify global configuration files (
~/.codex/AGENTS.mdand~/.claude/CLAUDE.md). This establishes persistent behavioral changes that apply across all future sessions for the agent. - [PROMPT_INJECTION]: The skill imposes strict constraints on the agent's operational logic and output format. Specifically, it mandates a blocking triage step when loading skills that requires a specific response format:
Triage <skill-name>: loading <files>; skipping <files> because <reason>.It also defines mandatory principles for decision-making and sub-agent management. - [EXTERNAL_DOWNLOADS]: The
THIRD-PARTY.mdfile contains links to various external skill repositories and tools. While some targets are well-known services (such as Upstash, Vercel, and Cloudflare), others are hosted by individual developers on GitHub. These are presented as recommended resources for the user to install. - [COMMAND_EXECUTION]: The skill provides instructions for managing
sudocommand execution by directing the agent to use a specific pane management skill (/herdr). This is designed to facilitate user password entry outside the standard agent context.
Audit Metadata