Files
SlpSoftware/aidlc-docs/_shared/assumptions-to-check.md
T
SluijsensandClaude Sonnet 5.5 d980cc72f8
Continuous Integration / config (pull_request) Successful in 17s
Continuous Integration / prepare (pull_request) Successful in 2m13s
Continuous Integration / build-production (pull_request) Skipped
Continuous Integration / test (pull_request) Successful in 3m14s
Continuous Integration / build (pull_request) Successful in 4m14s
Continuous Integration / deploy-test (pull_request) Skipped
Continuous Integration / lighthouse (pull_request) Successful in 3m19s
Continuous Integration / deploy-production (pull_request) Skipped
Hero (SLS): feature afgesloten voor test, aannamelijst en overzicht van aannames om te controleren
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-17 09:02:35 +01:00

2.8 KiB

Assumptions to check (coordination, SLS session)

Kept while the owner was away (2026-11-17). Each item: what I decided or assumed, why, what changes otherwise. The feature-specific lists live in each feature folder (CMS hero: 30 items, ASF hero: 10+, SLS hero: 9, media library: to follow).

# Assumption or decision Why If the owner decides differently
1 Blanket go (owner, 2026-11-17): when everything is ready, PRs are made and merged to master, and pushes and deploys go to TEST. No production deploys, no release/* branches, no branch-protection changes except what is needed to merge. Explicit message from the owner. Stop merging; the branches stay as they are.
2 Merge order: CMS first (hero-module, then messages-badge, then admin-form-controls), then ASF hero, then SLS hero. A merge needs green PR checks. The sites need the CMS host; the hero module is the largest change. Another order only matters for conflicts.
3 The nightly ASF test build is removed (PR #20 merged, efd2aee). No periodic dependency audit replaces it ("niet nodig nu", owner). The host injects CMS data per request (NFR-10); the nightly build overwrote test with master. Add a small weekly audit workflow, or Dependabot/Renovate.
4 After a merge to master, the master run deploys to test only. I redeploy test from the feature branches only as long as they are not in master. Production is manual and, for ASF and SLS, only from release/*. n/a
5 The CMS test deploys of both sites go in ONE run (owner, 2026-11-17). Fewer duplicate pipeline runs. Split them again if a health check fails because of simultaneous restarts.
6 Sub-agents: the CMS session works serially because its own rule says sub-agents need the owner's direct request in its session. I use Sonnet sub-agents for my own SLS work. Working agreement per session. The owner tells the CMS session to use sub-agents for parallel streams.
7 The LCP of about 5 s on SLS (mobile) is not part of the hero feature. Dominated by the app bundle; the hero is not the cause. A separate performance feature.
8 Media library: starts after the merges, requirements and design only; nothing that moves data or changes public media addresses is built without the owner. The owner chose "first the hero, then level 3"; data migrations and public addresses are risky. Allow building up front.
9 Privacy statements (SLS concept, ASF page) are not touched; they keep their [OPEN] items (Sentry retention and DPA, Umami, EU transfer, processing agreement with MijnHostingPartner, captcha paragraph). Owner kept them aside to review himself. n/a
10 The CMS session briefly used a stale local master; the Contact module is on master (PR #15, b2cacfe). Verified on origin/master. n/a