{
  "skill_name": "sop-gap-audit",
  "evals": [
    {
      "id": 1,
      "prompt": "I am the Operations Manager for customer onboarding at a B2B SaaS company. Audit this draft SOP before I publish it. Draft SOP: Sales marks enterprise deal closed-won in Salesforce. CSM creates onboarding project. Implementation schedules kickoff. Customer sends technical contacts and SSO requirements. Implementation configures workspace. CSM sends go-live email. Context: two recent customers went live before billing codes were set up, so invoices were delayed. Security review is required for SSO and SCIM, but the draft does not say who checks it. Legal says EU customers need a DPA attached before kickoff. Support keeps asking where to find the customer's admin contact. There is no standard step for customers who miss kickoff twice. Target is kickoff within 5 business days and go-live within 30 days. Tools are Salesforce, Asana, Zendesk, Stripe, and Google Drive.",
      "expected_output": "An Operations Manager SOP gap audit with a readiness rating below fully ready, gaps for billing-code setup, security-review ownership, EU DPA prerequisite, admin-contact system of record, missed-kickoff exception path, metric instrumentation, ownership matrix by role, prioritized backlog with acceptance criteria, and a rollout note."
    },
    {
      "id": 2,
      "prompt": "Please audit this stale SOP for monthly contractor access review. SOP says: HR exports active contractors from BambooHR. IT checks Okta accounts. Managers confirm access. IT disables accounts no longer needed. Context: internal audit found three contractors still had Jira access after their end date. HR exports are pulled on the first business day, but some contracts end mid-month. Managers often approve by Slack emoji. Finance owns the vendor roster but is not listed. No one knows where evidence is stored. Security wants review completion within 7 calendar days. IT says app owners, not IT, know whether GitHub and Notion access is still needed.",
      "expected_output": "A contractor-access SOP gap audit that flags timing mismatch, Slack emoji approval evidence weakness, missing Finance/vendor roster ownership, unclear evidence retention, app-owner confirmation gap, Security SLA measurement needs, and a readiness decision with must-fix controls before audit closure."
    },
    {
      "id": 3,
      "prompt": "I manage warehouse operations. Audit our SOP notes for damaged inbound inventory. Notes: Receiving team photographs damaged cartons, creates a warehouse ticket, moves items to quarantine shelf, then tells Purchasing. If vendor approves credit, Purchasing updates NetSuite. Problems: sometimes photos are missing scale/reference labels. Perishables cannot sit on the quarantine shelf longer than 4 hours. QA wants lot numbers captured for regulated SKUs. There is no weekend escalation path. Finance only learns about credits at month-end. Ops wants to reduce average resolution time from 6 days to 3 days. We use NetSuite, Jira, shared drive photos, and Slack.",
      "expected_output": "A warehouse SOP gap audit with process boundary, exception-path review for perishables/weekends/regulated SKUs, photo evidence standards, QA lot-number control, Finance credit visibility handoff, realistic resolution-time metrics, owner matrix, prioritized update backlog, and rollout/training recommendations."
    }
  ]
}