← all branches

feat/powerradar

needs attention
f34e4ce · incrementalPR #327reviewed 2026-07-23 00:32 UTC0H · 1M · 3L · 5I
The branch
Purpose
Add a native API-based PowerRadar (Centrica/Panoramic Power) metrics integration replacing the manual-CSV path with headless OAuth2 auth + 5/15/60-min pull collection via the internal /eigw API.
Goal
powerradar-api integration worker: OAuth broker auth, chart-data pull, translation layer, CDK deploy, live e2e proven on Energía Real.
Sub-goals
  • SG-1: Auth + transport engine (eigw-client.ts) — complete
  • SG-2: Manifest + catalog + provider registration — complete
  • SG-3: Pure translation + unit tests — complete
  • SG-4: Metrics handler + tests — complete
  • SG-5: CDK stack + wiring — complete
  • SG-6: Live worker e2e — complete
  • SG-7: Framework fold-back — complete
The changes (whole branch)
What
Merge of origin/main: (1) preview-provision.yml timeout 45→75 min with corrected slug-length comment (new longest CDK resource name is metrics-helioscope-materialize-persist); (2) pnpm-lock.yaml patch bumps: aws-cdk 2.1132.0→2.1132.1, tinyglobby 0.2.15→0.2.17, picomatch 4.0.3→4.0.4.
Why
Branch merged main to pick up CDK CLI bump (#335) and CFE silent-success fix (#336) landed while PowerRadar integration was being built.
Areas
services/metrics/integrations/powerradar-api+1450−0infra/cdk/src/stacks/services/metrics/integrations/powerradar-api+157−0infra/cdk/src/app + lib + engine+39−9packages/integration-manifests+109−0packages/database+17−1.github/workflows/preview-provision.yml+57−0
Blast
29 files, ~2318 insertions / ~360 deletions. New integration service, manifest, CDK stack, DB seed entries. No changes to existing integrations or public API surface.
New OAuth2 password-grant auth shape (headless, no MFA) Exploration handler deferred (metrics-only, solark precedent)
➖CI· No check results returned by GitHub API for this SHA➖coderabbit· No .coderabbit.yaml in repo

Findings · 10

correctness2

info

Timeout 45→75 min: well-justified, measured on fix/cfe-silent-ok (2026-07-22)

.github/workflows/preview-provision.yml:46

Comment correctly distinguishes first-provision (slow) vs incremental pushes (fast). No correctness concern.

info

Slug-length math verified correct

.github/workflows/preview-provision.yml:729

'batu-{slug}-dev-metrics-helioscope-materialize-persist' = 48 fixed chars → max slug = 16 chars. naming.ts MAX_BRANCH_NAME_LEN=16 matches. Hash-bounding behavior for 17-char slugs accurately described.

security2

info

aws-cdk 2.1132.0→2.1132.1: patch bump, integrity hash updated, no known CVEs

pnpm-lock.yaml

Consistent update in packages and snapshots sections.

info

tinyglobby 0.2.15→0.2.17 + picomatch 4.0.3→4.0.4: routine transitive patch bumps

pnpm-lock.yaml

No known CVEs; transitive deps of typescript-eslint and rolldown-plugin-node-polyfills.

conventions3

medium

Warning message 'exceeds 17 chars' is ambiguous — gate fires at ≥18

.github/workflows/preview-provision.yml:740

Gate is `-gt 17` (slugs ≥18 skip CDK); slugs of exactly 17 are hash-bounded and still deploy. Suggest: "Branch slug is 18+ chars — skipping CDK preview. Slugs ≤16 deploy verbatim; exactly 17 chars are hash-bounded and still deploy. Rename to ≤16 chars (per CLAUDE.md)."

low

Warning rename hint omits the 16-char budget from CLAUDE.md

.github/workflows/preview-provision.yml:740

The hint says 'something shorter (e.g. feat/short-name)' without stating the ≤16-char target. Operators hitting this should know the budget without consulting CLAUDE.md.

low

Dual thresholds (16 in naming.ts, 17 as gate) lack a one-line rationale

.github/workflows/preview-provision.yml:734

A reader may wonder why the gate is >17 rather than >16. One line: 'Gate is >17 (not >16) because naming.ts hash-bounds 17-char slugs to a valid 16-char resource name; >17 breaks the fixed-overhead budget even with hash-bounding.'

tests1

info

No test files in incremental diff

Changes are CI config + lockfile only. Core PowerRadar integration tests (13 translation + 18 handler) landed in prior reviewed commits.

improvement1

low

75-min timeout: add a note on `--concurrency 2` as the next lever if estate grows further

.github/workflows/preview-provision.yml:50

Evidence-based and appropriate now. If future stack count pushes past 75 min, `cdk deploy --concurrency 2` can parallelize unrelated stack groups. Worth noting so the next person knows the knob.

seo1

info

No apps/web/** files in incremental diff — auto-skip

SEO lens not applicable to this diff.

History · 15 commits

  1. f34e4ceneeds attentionincremental0H · 1M · 3L2026-07-23 00:32current
  2. 1bc8aedneeds attentionincremental0H · 2M · 3L2026-07-22 21:34
  3. 31e63bbneeds attentionincremental0H · 2M · 5L2026-07-22 20:53
  4. 482cb88safeincremental0H · 0M · 1L2026-07-22 17:56
  5. 9a73b79safeincremental0H · 0M · 0L2026-07-22 17:34
  6. 87f8298needs attentionincremental0H · 1M · 3L2026-07-22 16:59
  7. 1e80096safeincremental0H · 0M · 1L2026-07-22 16:42
  8. 6ed65b4safeincremental0H · 1M · 3L2026-07-22 00:00
  9. 74a1131needs attentionfull0H · 7M · 12L2026-07-21 19:06
  10. 563252bneeds attentionincremental4H · 9M · 8L2026-07-21 18:32
  11. beced58needs attentionincremental0H · 3M · 3L2026-07-21 01:13
  12. 7fa8684needs attentionincremental4H · 9M · 7L2026-07-20 22:56
  13. 230784fneeds attentionfull2H · 11M · 14L2026-07-10 00:15
  14. b876b54needs attentionincremental2H · 2M · 5L2026-07-08 04:44
  15. ec4847fneeds attentionfull2H · 11M · 8L2026-07-08 03:32