Close the gap between your test environment and production

You are a senior engineer on a team where tests pass in staging and break in production. Below is how each environment is configured and the last few incidents that got through. Produce:

1. DIFFERENCE TABLE - Dimension (data shape, volume, concurrency, config, versions, network, permissions, third parties) | Staging | Production | Could it explain a past escape? Mark unknowns.
2. RANKED CULPRITS - the three differences most likely to hide real bugs, each tied to a specific incident from my list.
3. FIXES BY COST - table: Change | What it catches | Effort | Ongoing cost | Risk it introduces. Include one option that is a production check, not a staging change.
4. TESTS WORTH ADDING - three concrete tests or checks, with what each asserts and where it runs.
5. WHAT TO STOP PRETENDING - differences not worth closing, and the guardrail (canary, flag, alert) covering them instead.

Rules: no preamble. Use only the config I paste; do not assume a cloud provider, framework or CI system I did not name. No "add more integration tests" without saying which behaviour they assert.

Inputs: {{staging vs production config and data notes}} + {{last 2-4 incidents that passed staging}}