Days 15–30 — Independent triage

You take the queue. You still ask questions; you do not wait to be told to capture a HAR.

Day 15–18 — Data, read-only

Read SQL and MongoDB and database clients.
On staging, run a tenant-scoped SELECT / find with LIMIT. Show your lead the statement timeout.
Fill the private templates if your org has a product: npm run docs:dev:private and docs/product/.

Check: You refused (in writing) at least one impulse to UPDATE.

Own SEV3/SEV4 technical tickets end to end.
File one bug using the evidence template. File one "not a bug" with the HAR proving 403 + business code.
Run remaining scenario labs you skipped.

Check: Engineering did not come back with "need request id / HAR / tenant" on your bugs.

Complete this list with artifacts (ticket links, redacted). A lead initials each line.

Severity chosen correctly on a SEV2-or-higher candidate (or correctly kept at SEV3)
Sanitized HAR captured or coached, then interpreted
401 vs 403 called correctly, with body/header evidence
GraphQL or { success:false } 200 identified if your product has it; otherwise a REST problem+json / ad-hoc body parsed
502/503/504 not used interchangeably
Read-only data lookup, tenant-scoped, pasted as counts/keys
Escalation note an engineer used without a follow-up "what was the URL?"
No secrets in customer thread or public channels
All four scenario labs discussed with a lead

If a box is empty, you are not late — you are not done. Stay on supervised tickets until it is true.

You are not finished with this book. Lookup mode starts now. When a code surprises you, add a sentence to the error page and run npm run qa.