SF Plugin Core Refs Assets Testv0.0.1

What to consider before installing: 1) Do not run any included scripts against production orgs or with real credentials until you review them. The docs instruct running Python scripts and 'sf' CLI commands that require Salesforce credentials and possibly private keys, but the skill declares no required credentials — that mismatch is suspicious. 2) Inspect the bundle locally first. Because there's no install step, the host may unpack assets. Open and read any scripts (e.g., query-org-metadata.py, mermaid_preview.py, deploy.sh) and search them for network endpoints, hard-coded URLs, and unexpected outbound requests before executing. 3) Treat detected prompt-injection text seriously. The SKILL.md contains phrases that could try to alter an LLM's behavior; ask the author to remove these or explain why they're present. 4) If you want to try the skill, use an isolated sandbox environment and least-privilege credentials (a throwaway org or a service account with minimal permissions). Prefer manual invocation rather than allowing autonomous agent execution until you're confident. 5) Ask the publisher for clarification: which external tools are required (python, sf CLI, node, etc.), which environment variables or credentials will be used, whether any included scripts make external network calls, and why prompt-injection phrases appear in SKILL.md. 6) If you lack the ability to audit code, consider not installing, or request a signed/verified release or an install spec that declares required tools and secrets and documents safe usage.