Skip to content

docs: overenskomst implementation status after clause-by-clause audit - #1679

Merged
renemadsen merged 1 commit into
stablefrom
docs/overenskomst-implementation-status
Aug 7, 2026
Merged

docs: overenskomst implementation status after clause-by-clause audit#1679
renemadsen merged 1 commit into
stablefrom
docs/overenskomst-implementation-status

Conversation

@renemadsen

Copy link
Copy Markdown
Member

Documentation only. Records the result of re-verifying every GLS-A / 3F preset against the primary agreement PDFs (obtained and text-extracted, not summarised).

Verdict: the presets are not true to the documents

All 8 previously-listed discrepancies confirmed; 8 further categories found.

Under-attribution (workers coded too low) — Jordbrug Elev o18 uses 30 % where § 47 stk. 4 gives lærlinge 50 %; all 14 "u18" presets are missing their top overtime tier, so hour 3 onward stays at the first-tier rate indefinitely; Agro Kartoffelsortering, Gulerod and Minkfoder each code overtime 20 points light; Skovbrug gives 6 h of "normal" Saturday when § 7 stk. 1 puts the week on "ugens 5 første hverdage".

Over-attribution — the Saturday-afternoon supplement is fabricated in Jordbrug Standard and all 8 Agro Standard presets (an exhaustive "lørdag" search finds it only for dyrehold, lærling stald and praktikant); Dyrehold's evening band runs 4 h past what § 23 allows; Skovbrug's evening band is 1 h too long.

Structural — Agro Øvrige models percentages where § 4 a pays flat DKK across clock bands; Agroindustri encodes no forskudt bands at all despite § 19 defining them; Grundlovsdag is a flat all-day code on 37 of 39 presets.

The Elev problem is bigger than a threshold: the u18/o18 split is fabricated. Each agreement has exactly one lærling overtime rule — or none. Agroindustri contains the word "lærling" zero times, yet 8 Agro Elev presets ship.

What is correct

The two Udenlandske praktikanter presets, as of the corrections merged today — tiers, Saturday, Sunday and the § 29 Grundlovsdag noon split all match.

Also recorded: why fixing these is not just a catalogue edit

  • Presets are copy-at-create-time snapshots; correcting one does not touch rule sets already created, and no migration exists. Today's regression proved the cost: a name-gated engine change met stale rows and moved up to 6 h/Saturday onto an afternoon supplement.
  • Weekday bands shadow the tiers on the 8 "Standard" presets, so several tier errors are currently invisible rather than harmless.
  • Pay lines are never persisted, so re-exporting an already-paid period reflects today's rules with no snapshot to reconcile against.

Ends with a recommended sequence, each step of which needs a migration story before it ships.

Verification caveat

Gartneri, Skovbrug and Golf presets claim 2026-2029 but only 2024-2026 texts are published — flagged in the doc.

🤖 Generated with Claude Code

…clause audit

Every GLS-A/3F preset re-verified against the primary agreement PDFs. The
presets are not true to the documents: all 8 previously-listed discrepancies
confirmed, 8 further categories found.

Worst are under-attributions — Jordbrug Elev o18 uses 30% where the text gives
lærlinge 50%, all 14 u18 presets are missing their top overtime tier, and
three Agro presets code overtime 20 points light. The u18/o18 split itself is
fabricated: each agreement has exactly one lærling overtime rule or none, and
Agroindustri never uses the word at all despite 8 Elev presets shipping.

Also records the over-attributions (fabricated Saturday supplements, band
overruns), the structural gaps (Agro Øvrige pays flat DKK not percentages; no
Agro forskudt bands; Grundlovsdag flat on 37 of 39 presets), and the
engine-level facts that decide what actually surfaces — bands shadowing tiers,
presets being copy-at-create-time snapshots with no migration, and pay lines
never being persisted.

Ends with a recommended sequence, each step of which needs a data migration
rather than only a catalogue edit.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Copilot AI lite review requested due to automatic review settings August 7, 2026 17:02
@renemadsen
renemadsen merged commit 273d8a0 into stable Aug 7, 2026
1 check passed
@renemadsen
renemadsen deleted the docs/overenskomst-implementation-status branch August 7, 2026 17:02

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Adds an “implementation status” documentation section capturing the 2026-08-07 clause-by-clause audit results for all GLS-A / 3F overenskomst presets, clarifying which preset families are currently correct vs. discrepant and outlining the highest-impact defect categories plus a proposed remediation sequence.

Changes:

  • Added a new audit section summarizing verified agreement texts vs. what presets claim, including publication caveats.
  • Documented preset-family status (correct vs. incorrect) and a prioritized defect list by payroll impact.
  • Recorded engine-level considerations and a recommended migration-first fixing sequence.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

| Agroindustri 4012 | **2026-2029** (endelig 07.07.26) | 2026-2029 ✅ |
| Gartneri 4011 | 2024-2026 (no 2026-2029 published) | 2026-2029 ⚠️ |
| Skovbrug 4013 | 2024-2026 (no 2026-2029 published) | 2026-2029 ⚠️ |
| Golf 4014 | 2024-2026 (no 2026-2029 exists) | 2026-2029 ⚠️ |

# IMPLEMENTATION STATUS — all overenskomst presets (audit 2026-08-07)

Every GLS-A / 3F preset was re-verified clause by clause against the primary
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants