Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed iOffice MCP integration with powerful workplace-management actions, but the behavior is coherent with its stated purpose and includes write-confirmation safeguards.
Install only if you are authorized to automate your employer's iOffice/Eptura tenant. Use a least-privileged token where possible, verify the npm package or source repository, and be cautious with delete, approval, check-in/out, delivery, and user-management actions because they affect real workplace records.
Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.
Add to .mcp.json in your project or ~/.claude/mcp.json:
{
The trigger text includes a broad catch-all condition: 'any request involving workplace management in iOffice.' That can cause the skill to activate on loosely related prompts and expose powerful read/write workplace-management actions, increasing the chance of unintended tool use on sensitive tenant data. The risk is amplified because the skill supports creation, update, deletion, check-in/out, approval, and delivery workflows across buildings, users, visitors, reservations, moves, and mail.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
The 18 read tools — every `io_list_*` and `io_get_*` across the nine resource
groups — take `view: "compact" | "full"`, and **`compact` is the default**: the
parameter is optional, and omitting it gives you the slim shape without asking
for it.
**Compact strips media; it does not project fields.** No hand-written field
No suspicious patterns detected.