Fix opendata-ci auto-commit gap and add ISF/eligibility PowerShell commands

[Michael Flanakin (flanakin)]

🛠️ Description

opendata-ci.yml used git commit -a after regenerating PowerShell functions from src/open-data/*.csv. git commit -a only stages modified/deleted tracked files — it silently skips new untracked files. InstanceSizeFlexibility.csv (#2199) and CommitmentDiscountEligibility.csv (#2251) never had their generated Get-OpenData*.ps1 functions committed, so they stayed untracked forever and every PR touching either CSV (most recently #2295) saw a "no diff" CI result that masked the standing gap. Fixed by switching to git add -A before git commit.

Also adds Get-FinOpsInstanceSizeFlexibility and Get-FinOpsCommitmentDiscountEligibility public cmdlets, matching the existing Get-FinOpsPricingUnit/Get-FinOpsService pattern — these two datasets had no PowerShell-facing lookup at all, only direct CSV consumption (Power BI, Optimization Engine).

Bugs found and fixed along the way

• Get-FinOpsCommitmentDiscountEligibility's initial draft used the same Select-Object -Property * -Unique post-filter as the other Get-FinOps* cmdlets. At 92k+ rows this hung for minutes. MeterId is already unique in the source data, so -Unique was dropped (both from the cmdlet and its test's BeforeAll).
• Build-OpenData.ps1's Write-Command/Out-File generation path silently dropped rows on large datasets on some runs (662 rows missing from one CommitmentDiscountEligibility generation, 1 row from an InstanceSizeFlexibility generation). Regenerating fixed it both times. Not root-caused further — flagging here as a latent reliability risk in the generator for anyone touching it next.
• Merging in dev picked up fix(open-data,hubs): align commitment discount eligibility with FOCUS spend and usage categories #2284, which swapped which eligibility column maps to reservations vs. savings plans. Corrected the new cmdlet's help text/docs to match and regenerated the private function against the updated CSV.

Fixes # (opendata-ci not running as expected on PRs like #2295)

📷 Screenshots

N/A — PowerShell module change

📋 Checklist

🔬 How did you test this change?

• 🤏 Lint tests
• 🤞 PS -WhatIf / az validate
• 👍 Manually deployed + verified
• 💪 Unit tests
• 🙌 Integration tests

📦 Deploy to test?

• Hubs + ADX (managed)
• Hubs + Fabric (manual) — URI:
• Hubs (manual)
• Hubs (no data)
• Workbooks
• Alerts

🙋‍♀️ Do any of the following that apply?

• 🚨 This is a breaking change.
• 🤏 The change is less than 20 lines of code.

📑 Did you update docs/changelog.md?

• ✅ Updated changelog (required for dev PRs)
• ➡️ Will add log in a future PR (feature branch PRs only)
• ❎ Log not needed (small/internal change)

📖 Did you update documentation?

• ✅ Public docs in docs (required for dev)
• ✅ Public docs in docs-mslearn (required for dev)
• ✅ Internal dev docs in docs-wiki (required for dev)
• ✅ Internal dev docs in src (required for dev)
• ➡️ Will add docs in a future PR (feature branch PRs only)
• ❎ Docs not needed (small/internal change)

🤖 [AI] Generated with Claude Code

[Copilot]

🟡 Changes recommended

A few newly added/edited docs strings are internally inconsistent with the cmdlet defaults and contain an incorrect command description that should be corrected before merge.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

This PR fixes the Open Data CI auto-commit behavior so newly generated PowerShell files are actually staged/committed, and adds new PowerShell cmdlets + docs/tests to expose the Instance Size Flexibility and Commitment Discount Eligibility datasets via the FinOpsToolkit module.

Changes:

• Update opendata-ci.yml to git add -A before committing generated files so untracked outputs are not silently skipped.
• Add Get-FinOpsInstanceSizeFlexibility and Get-FinOpsCommitmentDiscountEligibility public cmdlets (and corresponding unit tests).
• Add/refresh generated Get-OpenData* private functions + unit tests and wire up Microsoft Learn docs/TOC entries.

File summaries

File	Description
.github/workflows/opendata-ci.yml	Fixes CI auto-commit to include newly generated (previously untracked) files.
src/powershell/Public/Get-FinOpsInstanceSizeFlexibility.ps1	New public cmdlet for querying ISF open data with wildcard filters.
src/powershell/Public/Get-FinOpsCommitmentDiscountEligibility.ps1	New public cmdlet for querying commitment discount eligibility by meter ID and eligibility flags.
src/powershell/Private/Get-OpenDataInstanceSizeFlexibility.ps1	Generated private lookup function for ISF dataset.
src/powershell/Private/Get-OpenDataCommitmentDiscountEligibility.ps1	Generated private lookup function for commitment discount eligibility dataset.
src/powershell/Tests/Unit/Get-OpenDataInstanceSizeFlexibility.Tests.ps1	Unit test ensuring generated ISF function matches CSV row count.
src/powershell/Tests/Unit/Get-OpenDataCommitmentDiscountEligibility.Tests.ps1	Unit test ensuring generated eligibility function matches CSV row count.
src/powershell/Tests/Unit/Get-FinOpsInstanceSizeFlexibility.Tests.ps1	Unit tests for new public ISF cmdlet behavior and typing.
src/powershell/Tests/Unit/Get-FinOpsCommitmentDiscountEligibility.Tests.ps1	Unit tests for new public eligibility cmdlet behavior and filters.
docs-mslearn/toolkit/powershell/powershell-commands.md	Adds the new cmdlets to the PowerShell command reference list.
docs-mslearn/toolkit/powershell/data/open-data-commands.md	Adds the new cmdlets to the open data command list.
docs-mslearn/toolkit/powershell/data/get-finopsinstancesizeflexibility.md	New Microsoft Learn reference page for the ISF cmdlet.
docs-mslearn/toolkit/powershell/data/get-finopscommitmentdiscounteligibility.md	New Microsoft Learn reference page for the eligibility cmdlet.
docs-mslearn/toolkit/open-data.md	Adds “See PowerShell” links for the two datasets.
docs-mslearn/TOC.yml	Adds TOC entries for the new cmdlet docs pages.

Review details

Suppressed comments (1)

src/powershell/Public/Get-FinOpsCommitmentDiscountEligibility.ps1:16

• The comment-based help for UsageEligibility says the default is null, but the parameter default is "*" and the filter uses -like. Update the help text so it matches the actual default behavior.

    .PARAMETER UsageEligibility
    Optional. Filters to meters with the specified reservation (usage commitment) eligibility. FOCUS classifies a reservation as a usage commitment because you commit to a quantity of usage. Expected values: Eligible, Not Eligible. Default = null (all).

• Files reviewed: 13/15 changed files
• Comments generated: 4
• Review effort level: Lite

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

[Roland Krummenacher (RolandKrummenacher)]

Thanks for splitting the CI fix out — git add -A is the right call, and I confirmed the old git commit -a genuinely skipped newly generated (untracked) files. The generated data itself checks out: both private functions match their CSVs exactly (92,624 and 1,353 rows, field for field), so the "generator dropped rows" hazard isn't present in this output. Eligibility semantics line up with the post-#2284 orientation, and the 6 new tests pass locally under Pester 6.

Requesting changes on one thing that's much cheaper to decide now than after this ships, plus a small correctness bug (inline).

The 16.2 MB generated file is the one I'd like to settle before merge. GitHub won't let me anchor a comment there (no diff entry for a file that size), so it goes here.

src/powershell/Private/Get-OpenDataCommitmentDiscountEligibility.ps1 is 16.2 MB / 92,633 lines, and FinOpsToolkit.psm1:7-17 dot-sources every Private/*.ps1 eagerly on import. Measured cold import: 0.72s without this file, 3.6-4.8s with it — every user of every cmdlet pays that, and Build-PsModule.ps1:83 ships the whole Private folder to PSGallery.

Second cost: Tests/Lint/Lint.Tests.ps1:43-44 runs Invoke-ScriptAnalyzer -IncludeRule $_ once per rule per file. I measured 10.9s for a single rule on this file, against 1.57s for the previously largest generated file. At ~75 rules that's roughly +14 minutes of lint on every src/powershell/** PR locally, and more on the Windows CI runner.

Embedding generated data as a function is the established pattern here, so this isn't a wrong approach — it's that this file is about 10x larger than anything we've embedded before, while the source CSV is only 5.5 MB. Import-Csv on demand, or lazy-loading behind a script-scoped cache, would avoid both costs. Happy to be overruled if the data has to be in-module, but I'd rather make that call now than after v15 is on the gallery.

Two more that can't be anchored because the files aren't in this diff:

src/templates/agent-plugin/skills/finops-toolkit/SKILL.md — the "Open data commands" table still lists only the five existing references. The plugin bundles the two new reference docs but never points at them, and no test enforces that table, so it will stay stale silently.

docs/open-data.md:40 — the Jekyll instance size flexibility tile keeps its hidden placeholder button (visibility:hidden) where every other dataset tile has a "See PowerShell" link. The docs-mslearn twin was updated in this PR, so the two sites now disagree on whether a cmdlet exists.

[Roland Krummenacher (RolandKrummenacher)]

Confirmed this is the actual gap: git commit -a stages only modifications to already-tracked files, so newly generated open-data files were never committed. Worth noting the failure was silent because a failing git commit is followed by git push, which exits 0 — the job stayed green while doing nothing. Might be worth a set -e or an explicit check later, but not in this PR.

[Michael Flanakin (flanakin)]

🤖 [AI] 👍 Acknowledged — thanks for confirming the root cause. Agreed a set -e or explicit exit-code check would help catch this class of failure earlier; leaving that for a separate follow-up as you suggested.

[microsoft-github-policy-service]

@Michael Flanakin (@flanakin): you have some new feedback!

Please review and resolve all comments and I'll let reviewers know by removing the Needs: Attention label. If I miss anything, just reply with #needs-review and I'll update the status.

[Roland Krummenacher (RolandKrummenacher)]

Heads-up on an interaction with #2308, which grows src/open-data/InstanceSizeFlexibility.csv from 1,353 to 2,573 rows (restoring retired SKUs the Catalogs API no longer returns).

Get-OpenDataInstanceSizeFlexibility.ps1 in this PR is generated from that CSV at its current 1,353 rows, so the merge order matters:

• This PR first — the file becomes tracked on dev, and I regenerate it on [Open Data] Fix ISF ratio scale, region coverage and culture handling, and keep retired-SKU reservations visible #2308 before merging. No action for you.
• [Open Data] Fix ISF ratio scale, region coverage and culture handling, and keep retired-SKU reservations visible #2308 first — this branch would carry a generated file built from the old CSV. Open Data CI would not catch it either, because its paths: src/open-data/*.csv filter does not match this PR.

So merging this one first is the cleaner order, and I will handle the regeneration on my side.

Worth noting for the auto-commit fix itself: Open Data CI ran green on #2308 (head 77949cd8) and generated nothing, because git commit -a skipped the still-untracked ISF file — exactly the gap git add -A closes here.

[Roland Krummenacher (RolandKrummenacher)]

Update on the interaction I flagged earlier: still there, but much smaller than it was.

#2308 no longer backfills retired SKUs — that part was dropped. It now only widens the region sweep, so the CSV goes from 1,353 to 1,363 rows rather than 2,573. Get-OpenDataInstanceSizeFlexibility.ps1 here is generated at 1,353, so it would still be stale by 10 rows.

Same mechanism as before, so merging this one first remains the cleaner order and I will regenerate on my side afterwards. If #2308 lands first instead, note that Open Data CI would not catch it: its paths: src/open-data/*.csv filter does not match this PR.

[Michael Flanakin (flanakin)]

🤖 [AI] Addressing the review-body items that couldn't be anchored to a diff line:

SKILL.md "Open data commands" table — ✅ Implemented. Added the two missing rows (get-finopscommitmentdiscounteligibility.md, get-finopsinstancesizeflexibility.md); both docs were already bundled under references/, just unlisted.

docs/open-data.md:40 (ISF tile missing "See PowerShell") — ✅ Implemented. Added the link, matching the pattern used by the Pricing units / Regions / Services tiles.

16.2 MB Get-OpenDataCommitmentDiscountEligibility.ps1 (import time + lint cost) — 🤔 Needs discussion, but agreed and planned. You're right that this isn't a wrong approach, just wrong at this scale — we don't want to lazy-load just this one dataset and leave the other five inconsistent. Rather than block this PR on a design-relayout, I've scoped a follow-up PR that reworks all six Get-OpenData* functions (and Build-OpenData.ps1, the generator) to lazy-load via Import-Csv with a script-scoped cache, replacing the embedded-literal pattern everywhere at once. Will land as its own PR referencing this thread.

Roland Krummenacher (@RolandKrummenacher) — also thanks for the heads-up on the #2308 merge-order interaction; agreed this PR should land first and appreciate you handling the regeneration on your side.

[Michael Flanakin (flanakin)]

🤖 [AI] PR Update Summary

Addressed: 8 thread(s)/comment(s)

• ✅ Implemented: 6
• 🤔 Needs discussion: 1 (16.2 MB generated file — agreed, scoped as a separate follow-up PR reworking all six Get-OpenData* functions + the generator to lazy-load)
• 👍 Acknowledged: 1 (silent-failure root cause confirmation)

Key changes: Fixed the SpendEligibility/UsageEligibility explicit-$null bug (was silently returning 0 rows), corrected copy-pasted/stale docs (Get-FinOpsPricingUnit description, eligibility parameter defaults, SKILL.md table, ISF tile link), and dropped a no-op -Unique call from Get-FinOpsInstanceSizeFlexibility. All 10 relevant Pester tests pass; PSScriptAnalyzer clean on changed files.