Security risk report for @forgecat/openai_skills_render-deploy v0.0.9
Source Integrity
Low
Profile is sourced from OpenAI's official skills repository (https://github.com/openai/skills) with Apache-2.0 license.
Metadata indicates original commit hash and timestamp; version 0.0.9 is tracked.
No supply-chain red flags; dependency list is empty and MCP server points to official Render endpoint.
Agent Intent
Low
Content describes legitimate deployment workflows and best practices; no instructions to ignore system prompts, exfiltrate credentials, or install malicious payloads.
References to reading environment variables (e.g., checking PORT, DATABASE_URL) are standard deployment patterns, not credential theft.
Documentation is educational and prescriptive for Render deployment only; no hidden guidance to poison future outputs or weaken security defaults.
Details
Evidence
Example: 'Bind to environment PORT' and 'Check if user is logged in' are normal operational steps, not manipulation.
No paraphrased or obfuscated instructions to bypass security, leak system prompts, or execute remote code.
Guidance on MCP setup and API key handling is transparent and user-initiated, not coercive.
Permissions
PRM-000Medium
The MCP server at https://mcp.render.com/mcp provides broad cloud infrastructure management capabilities including creating/deleting services, databases, and updating environment variables, which is wider than a read-only deployment assistant.
The skill explicitly guides the agent to call tools like create_postgres(), update_environment_variables(), and list_logs() — these are write/mutate operations on live cloud infrastructure.
However, the authority is consistent with the stated purpose (deploying to Render), and no unrelated high-risk tool categories like shell execution or local filesystem writes are declared.
The MCP server points to https://mcp.render.com/mcp, which is Render's official MCP endpoint — a known, legitimate service rather than an unknown or attacker-controlled binary.
The server requires a RENDER_API_KEY credential, meaning any compromise of that key grants the MCP server full access to the user's Render account including all services and databases.
No hidden instructions were found in tool descriptions, and the server URL is a well-known external host with scoped access to Render's API only — not unrestricted filesystem or network access.
ForgeCat