Skip to content
Compliance for Technical Staff

Lesson 4: Make changes you can explain

You will be able to: review a change for common security defects, and identify the evidence that shows the change followed the process.

About 4 min

Change management means changes are authorized, designed, developed, tested, approved, and deployed in a controlled way, so the organization can show what changed, who approved it, and why. For most teams the pull request is where this happens: a reviewer other than the author, required checks, and a deployment record that ties back to the merge.

Source: AICPA Trust Services Criteria, CC8.1(opens in a new tab)

Secure development builds on the same workflow: secure design, code review that looks for security defects, automated tests, dependency and vulnerability scanning, and secret scanning.

Source: NIST SP 800-218 (SSDF)(opens in a new tab)

  • Vulnerabilities are fixed within timeframes the organization sets according to risk. Frameworks generally require a defined, risk-based process; they do not set one universal deadline.
  • Exceptions happen. When a requirement cannot be met on time, record a time-bound exception that an accountable owner approves, with compensating controls and a remediation date.
  • Emergency changes can use an expedited path, but still need approval and a review after the fact.

Exercise 1 of 2: Review the pull request

A teammate opened a pull request that adds an invoice endpoint to Brightwell’s billing service. Each customer account should only ever see its own invoices.

Flag each added line that contains a defect and choose the type of defect. Then choose fixes and the evidence that would show this change followed the process.

src/routes/invoices.ts0 lines flagged
  1. 1import { Router } from "express";
  2. 2import { db } from "../db";
  3. 3import { logger } from "../logger";
  4. 4import { requireAuth } from "../auth";
  5. 5
  6. const LEDGER_API_KEY = "lk_live_7f3a9c2e41b8d6f05a1e";
  7. 7
  8. 8export const invoices = Router();
  9. 9
  10. invoices.get("/invoices/:id", requireAuth, async (req, res) => {
  11. logger.info("invoice request", { user: req.user, headers: req.headers });
  12. const invoice = await db.invoices.findById(req.params.id);
  13. if (!invoice) return res.status(404).end();
  14. const ledger = await fetch(`https://ledger.example/v1/balances/${invoice.accountId}`, {
  15. headers: { Authorization: `Bearer ${LEDGER_API_KEY}` },
  16. });
  17. res.json({ id: invoice.id, total: invoice.total, balance: (await ledger.json()).balance });
  18. });

Select a line number to flag it, then choose the type of defect. Unflagged lines count as “no defect.”

Fix for the hard-coded secret
Fix for the log line
Fix for the invoice lookup
Which records best show this change followed the process? Select all that apply.

Apply it 2 of 2: A fix you cannot ship this week

A critical vulnerability is published for a library you use. A patched version exists, but upgrading breaks a feature you cannot fix before your organization’s remediation deadline.

What is the compliant path?

Choose the best option.

Key points

  • Every production change should be explainable: who changed it, who approved it, how it was tested, when it shipped.
  • In review, look for secrets in code, sensitive data in logs, and missing authorization checks.
  • Authentication says who the caller is. Authorization says whether they may do this. Check both on the server.
  • When you cannot meet a deadline, get a time-bound, approved exception with compensating controls. Never hide the finding.