Your client: Halden · week 5 of 10
The extract changed overnight
Your client: Halden · Pro
Weeks 2 to 10 are in Pro.
Week 1 is free. Pro opens each later week’s brief, inbox, graded checks and deliverables, and records every grade here as your Halden record.
- Every stage and specialisation, in full
- Every lab and every role project
- Halden weeks 2–10, on your record
- Every drill, live in the playground
- All solutions, plus the server judge
- Lab Assistant and every Teach with AI
How a Halden week works
Each week opens on a Monday at Halden. Open the week's workspace and everything happens here, in your browser: no install and no terminal.
- Inbox. Read the week's emails; their attachments open beside them.
- Task. What happened, your tasks, and what done looks like.
- Files. One repository for the whole residency. Edit any file; Python and JSON are checked as you type.
- Run. Run pipeline rebuilds
out/from the latest pack (from week 2), Run a script runs one of your own modules, and Run tests runstests/. - Grade week. Runs this week's checks. When every required check passes, Start week N+1 opens the next one.
Week 1 is free and starts with seed 2026. Your repository stays in this browser; download it from the workspace whenever you want a copy. Keeping a copy at the end of each week is a good habit: the history is part of what you hand over.
The starter is a repository with an interface, some given code and TODOs.
Path Given Yours pipelines/clean.pya pass-through that accounts for every row the cleaning (week 2), schema versions (week 5) halden_app/model.pythe store: values with their source and history build()(week 3)halden_app/actions.pythe users and skills submit()and the four actions (week 3)halden_app/api.pyload(),model(), and the contract in the docstringclassify(),draft()(week 4),handle()(week 8)halden_app/sites.pythe product's site logic, complete nothing: from week 9 you configure it, you don't change it evals/an empty golden set and a judge stub both (week 4) config/users, the Santos site, the v1 mapping mappings, routing, Long Beach (weeks 5, 9, 10) docs/templates for week 1, and an index every document tests/a few shape checks your tests The interface the grader calls:
Call Returns From api.load(out_dir)the cleaned tables your pipeline wrote week 3 api.model(tables)the operational model; the grader reads it through m.val(key, prop)week 3 actions.submit(m, name, args, actor, key, now, approvers){"status": "applied" | "refused" | "pending_approval", ...}, auditedweek 3 api.classify(m, exception_id){"type", "cause", "action", "flags", "usd"}week 4 api.draft(m, exception_id){"to", "language", "subject", "body", "needs_approval", "usd"}, orNoneweek 4 judge(draft, case)inevals/judge.py"pass"or"fail"week 4 api.handle(m, exception_id, portal, now){"outcome": "drafted" | "routed", "reason", "key", "eta"}week 8 sites.load(name),sites.decide(case, cfg)the product's; the grader feeds your configuration through them week 9 A deterministic path passes every check. The sandbox has no network, so your code makes no model calls here.
Person Role First writes VP of operations Halden's sponsor; decides what goes live week 1 Rafaela Souza São Paulo desk lead; moves to Rotterdam on 11 May week 1 Halden IT service desk your access requests week 1 SI account lead Halden's systems integrator; sends the data week 2 Priya, central operations reports to the COO; owns right-first-time week 3 FinOps the budget week 4 Chief Information Security Officer security findings, and what may leave the building week 6 Camila Rocha acting São Paulo desk lead from 11 May week 7 Dewi Lestari Jakarta desk lead week 7 Halden on-call alerting the exception desk's SLO alert, not a person week 8 Maria Delgado Long Beach desk lead week 9 Product lead, your company owns sites.pyand the productweek 9 Priya Menon the SI's delivery lead; takes over on 12 June (not Priya in central operations) week 10