chore: track Dashboard API 1.74 and add the model gap diff

[danielabbatt]

What

Two things, no library code changes:

1. version.json moves from 1.70 to 1.74. v1.74.0 is the latest stable tag in meraki/openapi, published 2026-09-02 (v1.73.0 was 2026-08-05). This resets the Nerdbank.GitVersioning height, so builds jump from 1.70.138 to 1.74.x rather than continuing to 1.70.139. The 1.70.138 section stays in the changelog as history — it is accurate for what main built as at 84a0c39b.

2. A repeatable model gap diff, .github/skills/meraki-api-update/Find-MissingModelMembers.ps1, plus its checked-in output gap-report-v1.74.0.md.

The bump asserts which spec version the library is measured against. It does not claim every 1.74 endpoint is implemented — the report below is exactly how far off that is.

How the diff works

It reflects over the compiled Refit interfaces, matches each method's verb and path template to a spec operation, then walks every 2xx JSON response schema in parallel with the C# return type, comparing spec property names against [DataMember] names.

Three classes of legitimate mismatch are excluded, because the spec describes them as ordinary properties while the models correctly represent them as open-ended maps or as no payload:

• Dictionary-typed models — HTTP status-code maps, per-camera-model video settings.
• Methods declared to return the non-generic Task.
• Purely numeric property names — latency histogram buckets, HTTP status codes.

Each of those was a false positive observed in a real run before being filtered; without them the count is 234 rather than 155.

What it found against v1.74.0

Measure	Count
Refit endpoints in the library	1075
Operations in the v1.74.0 spec	998
Library endpoints matched to a spec operation	861
Library endpoints with no spec match	214
Spec operations not implemented	140
Unmapped response members on existing models	155, across 86 types

118 of the 155 are scalars; 37 objects and 23 arrays need roughly 57 new nested classes alongside them.

Area	Members	Of which object/array
Organizations	43	22
SM	27	5
Appliance	25	13
Wireless	19	6
General	10	1
Camera	8	1
CellularGateway	7	1
Switch	7	2
Sensor	6	5
LiveTools	2	0
Licensing	1	1

Spot-checked and real: SwitchPort.stpPortFastTrunk, Ssid.campusGateway and .wlanIdentifier, Vlan.sgt, Client.cdp, and SmVppAccount missing 13 of its fields.

Reproduce

pwsh -File .\.github\skills\meraki-api-update\Prepare-MerakiApiUpdate.ps1 -Version 1.74.0
dotnet build Meraki.Api.slnx -c Debug
pwsh -File .\.github\skills\meraki-api-update\Find-MissingModelMembers.ps1 -SpecVersion v1.74.0 -OutCsv gap.csv

Verification

dotnet build Meraki.Api.slnx -c Debug   Build succeeded, 0 Error(s), 6 pre-existing CS0618 warnings
Version of the branch head              1.74.4 (main will be 1.74.5 after a merge commit)
Meraki.Api.Test (Data namespace)        Total: 14, Errors: 0, Failed: 0, Skipped: 0
Meraki.Api.Test (Workflows)             Total: 12, Errors: 0, Failed: 0, Skipped: 0

Merge with a merge commit for the 1.74.5 changelog label to be correct; a squash makes it 1.74.4.

Follow-up

The 155 members land as one PR per product area, largest first, each with its own changelog entry and regression tests.

Notes

• The 155 is the accumulated gap, not only what 1.71 → 1.74 added. No attempt was made to attribute each member to the release that introduced it.
• The 214 unmatched library endpoints are not necessarily wrong — some are endpoints Cisco removed or renamed, some are path shapes the normaliser does not line up. They need their own pass; a few may be dead code.
• SKILL.md now tells the reader to run the diff rather than read the changelog narratively, and warns that a full run finds more than one release introduced.
• Cisco's What's New page still lists 1.73 as current, while the spec repo has tagged v1.74.0 — flagging in case 1.74 is not meant to be public yet.

[codacy-production]

Not up to standards ⛔

🔴 Issues 2 medium

Alerts:
⚠ 2 issues (≤ 0 issues of at least minor severity)

Results:
2 new issues

Category	Results
ErrorProne	2 medium

View in Codacy

AI Reviewer: run a review on demand. To trigger the first review automatically, go to your organization or repository integration settings. AI can make mistakes. Always validate suggestions.

_{TIP This summary will be updated as you push new changes.}