Security audit
Security checks for vulnerabilities and agentic risk
This skill uses a disclosed shell wrapper to query OpenAI documentation and does not show hidden persistence, credential access, destructive behavior, or unrelated data handling.
Install this if you want agents to verify OpenAI API and SDK answers against current official documentation. Be mindful that search terms and fetched documentation URLs are sent to the configured MCP endpoint, and avoid placing secrets or private customer data in documentation queries.
The skill invokes shell-based capabilities through a CLI wrapper but does not declare any tool scope restrictions such as allowed tools or permissions. This creates an unnecessary expansion of execution capability and makes it harder for an agent runtime or reviewer to enforce least privilege, especially if the script or environment is modified later.
The description says to use this skill whenever a task involves the OpenAI API or related areas, which is broad enough to trigger on many loosely related tasks. Overbroad activation can cause unnecessary shell/network use and may route sensitive task context into external queries when a narrower, local answer would suffice.
The instruction to 'Always use this skill' lacks boundaries and encourages unconditional use for OpenAI-related work. In practice, this can override safer decision-making, causing needless external access and broader-than-necessary shell execution, particularly in contexts involving confidential prompts or data.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
mcp_request() {
local payload="$1"
curl -s -N \
-H 'Accept: application/json, text/event-stream' \
-H 'Content-Type: application/json' \
-X POST "$MCP_URL" \
No suspicious patterns detected.