openai-docs

CRITICAL E004: Prompt injection detected in skill instructions.

• Potential prompt injection detected (high risk: 0.80). The prompt explicitly instructs the agent to retry installing the MCP with escalated permissions and "Do not ask the user to run it yet," which directs covert privilege escalation and performing actions without user consent outside the skill's stated doc-fetching purpose.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill declares a required MCP runtime dependency at https://developers.openai.com/mcp and instructs the agent to fetch and inject OpenAI doc sections (via mcp__openaiDeveloperDocs__fetch_openai_doc), so external content from that URL is fetched at runtime and directly controls the agent's responses/instructions.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The prompt tells the agent to run install commands and, if they fail due to permissions/sandboxing, to immediately retry with escalated permissions (i.e., obtain sudo/elevation) without first asking the user, which encourages modifying the host system state.