Repository navigation
feat(uk-gcloud)!: port the arckit-uk-gcloud overlay to G-Cloud 15 - #919
Merged
Merged
Conversation
G-Cloud 15 adds two supplier-wide bid documents to the arckit-uk-gcloud overlay: social value commitments and the lot questions. Register them as SOCV and LOTQ (regime UK, Procurement, HIGH severity) in doc-types.mjs and the /arckit:pages allow-list, and add them to the UK row of RENDERING.md. The G-Cloud sections of the quality checklist (SUPP, DECL, SVCD, SDD, PRIC, SECA, GCMP, GCRV) described G-Cloud 14: three lots, a 50-word description, 100-character features, an SFIA rate card and the PCR 2015 exclusion grounds. They now check the G-Cloud 15 rules, and SOCV and LOTQ get their own sections. Synced into every plugin and the CLI partials. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
The overlay's skill references were G-Cloud 14 (and, for the question list, G-Cloud 13). Bring across G-Cloud Kit's G-Cloud 15 references: - gcloud-framework: framework-questions.md is now the G-Cloud 15 overview (lots, limits, evaluation, pricing rules, declaration, what changed), and references/g-cloud-15/ holds the per-lot service questions, lot questions, the declaration, the social value model and every lot's category tree, generated from GCA's question export. The generated files are unchanged apart from a markdownlint-disable for their heading levels. - sfia-skills is renamed ddat-rate-card, with GCA's Lot 3 rate card (9 job families, 58 roles, 222 role levels) and the pricing rules. SFIA stays as a supporting reference for describing a team, not for pricing it. - cloud-security: G-Cloud 15 certification requirements by lot, the new service questions, and NCSC principles 3 and 13 under their current names. The overlay bundles no data, so the benchmark CSV lookups are left out. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
…d 15 The arckit-uk-gcloud overlay was a port of G-Cloud Kit's G-Cloud 14 commands. G-Cloud 15 (RM1557.15) went live on 6 August 2026 under the Procurement Act 2023, run by the Government Commercial Agency (GCA, formerly CCS), and the Digital Marketplace lists nothing else. Port G-Cloud Kit's G-Cloud 15 rewrite (v0.8.0), keeping the overlay's ArcKit projects, document IDs, Document Control, citations and quality checks: - service-design chooses one of five lots (1a, 1b, 2a, 2b, 3), records it on a **G-Cloud Lot** line and names the matching SDD command. - sdd-lot1 and sdd-lot2 become sdd-lot1a/1b and sdd-lot2a/2b; sdd-lot3 is rewritten around the DDaT rate card. Templates are rebuilt from GCA's question export, and lots with the same questions share one. - New social-value (SOCV) and lot-questions (LOTQ) commands. The lot questions are one supplier-wide document with a Part per lot group. - declaration follows the Procurement Act 2023 (CDP, PPON, exclusions on the CDP, debarment list, supply chain payments) and never answers the bid-writer question for the supplier. - pricing, security, review, submission-pack, supplier-profile and gcloud-competitors follow each lot's G-Cloud 15 rules and limits. - The recipe adds SOCV and LOTQ and defaults to sdd-lot2b. BREAKING CHANGE: /arckit:sdd-lot1 and /arckit:sdd-lot2 are removed; use sdd-lot1a or sdd-lot1b, and sdd-lot2a or sdd-lot2b. The sfia-skills skill is now ddat-rate-card. G-Cloud 14 is no longer supported. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
- Guides: new sdd-lot1a, sdd-lot1b, sdd-lot2a, sdd-lot2b, social-value and lot-questions guides; sdd-lot1 and sdd-lot2 removed; the other nine overlay guides rewritten for G-Cloud 15. Synced to the plugin guide tree. - guides.html, commands.html, docs/README.md, llms.txt, manifest.json and guide-groups.mjs list the 15 commands; index.html and the root README describe the overlay as G-Cloud 15 with 15 commands. - DEPENDENCY-MATRIX.md gets a changelog entry with the new supplier path. - Both changelogs record the change under Unreleased. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Regenerated with scripts/sync-claude-plugin-layout.py. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
…ell zsh-safe Ports two G-Cloud Kit v0.8.2 fixes (8f46d1d, f6a5b6b). service-design ran create-project.sh straight away and only then said to reuse an existing project, but the helper always allocates a new number, so a re-run under another name left two projects for one service. It now lists the projects first, matches the input by number, name or path, and asks when an existing project looks like the same offer. Claude Code runs a command's shell in the user's shell, zsh on macOS, where a path glob that matches nothing aborts the block with "no matches found". service-design, supplier-profile, submission-pack, lot-questions and declaration listed documents with bare globs such as `ls projects/000-global/supplier/ARC-000-SOCV-v*.md`, which failed in exactly the projects that lack the document. They now use find, and a test rejects an unquoted path glob in any of the overlay's shell blocks. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Ports G-Cloud Kit v0.8.2's placeholder check (d17f4d5). review and submission-pack looked only for a bare `[PENDING]`, so the `[PENDING: …]` and `[PENDING — …]` forms the overlay's own commands write passed as finished, and so did older markers and template fields nobody filled in. G-Cloud Kit moved the rule into plugin/scripts/placeholders.py. ArcKit overlays ship no scripts of their own (every `scripts/` path an overlay command names is the core plugin's), so both commands now run the same inline awk scan instead: `[PENDING]` in every form, `[TODO]`, `[TBD]`, `[TBC]`, `[CONFIRM]`, `*[TO BE ADDED]*` and the like, and any field the overlay's templates use. Like the script it skips code spans, links, ticks and comments, and it also skips the Document Control Reviewed By and Approved By rows and the Revision History, which ArcKit leaves `[PENDING]` until approval and which are not bid answers. A test keeps the two copies identical and runs the scan on a sample document. The GCRV checklist and both guides describe the same rule. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Ports G-Cloud Kit v0.8.2's category fix (3eff0a6). The overlay let a service tick categories across roots and second-level groups. None of the 42,893 live G-Cloud 15 listings scraped on 7 October 2026 has categories in more than one Root > Group, though GCA's question export states no rule, so the listings are the evidence. The service design and every SDD template now record the one category group on a Category group line. service-design asks which group a listing covers and designs an offer that spans groups as one service per group, the five SDD commands enforce it, and review blocks on a service whose categories span groups. The framework reference and skill state the rule with its source, and the SVCD and SDD quality checks and the guides say the same. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Ports G-Cloud Kit v0.8.2's word limits (651824d, e974960). GCA's question export states word limits only for features, benefits, system requirements, backed-up items and the Lot 1a/1b scored answers, but the Digital Platform enforces a 50, 100 or 200-word limit on the other free-text questions. framework-questions.md now carries G-Cloud Kit's per-lot table of those limits, inferred from the 42,893 live listings with the evidence for each (580 Lot 2b answers to "Information security policies and processes" sit within 2 words of 200, for example). The SDD templates show each limit on a `**Words:** [X]/N` line, and the lot questions and pricing templates state theirs. G-Cloud Kit recounts the answers with plugin/scripts/word-limits.py; the overlay ships no scripts, so the five SDD commands and review run the same count as inline awk: each SDD command recounts what it wrote and rewrites anything over, and review reports every answer over its limit as blocking, a counter that disagrees with the recount, and an SDD written before the counters existed. Tests check every limit in the table reaches its template, keep the six copies of the count identical, and run it on a sample SDD and the templates. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Ports G-Cloud Kit v0.8.2's rate card fix (40a9c19). The SDD, service design and DDaT skill told suppliers to offer only the role levels each service needs, while pricing said one card covers all Lot 3 services. The listings settle it: every Lot 3 listing shows the supplier's whole card, only 2 of the 1,135 suppliers with more than one live Lot 3 service show different cards, and the median Lot 3 service shows all 222 role levels. sdd-lot3 also copied its rates from the pricing document while pricing took its levels from the SDD, a circle with no starting point. The card is now one supplier-wide document, registered as a new doc-type, RATE (G-Cloud Lot 3 Rate Card, regime UK, Procurement): pricing writes it once to projects/000-global/supplier/ARC-000-RATE-v*.md from the new rate-card-template.md, and each Lot 3 pricing document summarises it. sdd-lot3 lists the role levels that deliver the service and checks they are on the card without copying any rates. Roles outside DDaT (procurement and commercial advisers, trainers) are priced at the nearest DDaT role and level, recorded on the card and stated in the service definition document. review checks the card, submission-pack bundles it for a Lot 3 service, and the placeholder scan covers it. RATE is registered in doc-types.mjs, the /arckit:pages allow-list, the RENDERING.md regime index (synced to every plugin and the CLI partial), docs/manifest.json and the quality checklist, which gains a RATE section. G-Cloud Kit's per-level benchmark columns are not ported: the overlay bundles no benchmark data, so the card compares rates with the DDaT skill's market table or rival listings, as the PR's pricing already did. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
…l-off warning Ports G-Cloud Kit v0.8.2's Cyber Essentials fix (b18e5a0). lot-questions, security and review listed Cyber Essentials (Lots 2a, 2b and 3) and Cyber Essentials Plus (Lots 1a/1b) under "would fail". Framework Schedule 1 v2.1 makes them mandatory for call-off contracts, Attachment 2 v5.0 doesn't make them conditions of the tender, and 769 Lot 2b and 1,977 Lot 3 listings are live with "Cyber essentials: No" and "None of the criteria". The commands, templates, skills, guides and quality checklist now report a missing certificate as a call-off warning, and lot-questions has a Call-off Warnings section beside Would Fail as Written. The alternatives use the live listings' exact wording: on Lots 2a, 2b and 3 they say "within 12 months of the date of award", where the overlay (following the question export) said "by the date of framework award". Lot 1a listings keep the export's wording. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
…us-framework designs Ports G-Cloud Kit v0.8.2's workflow order fix (ddf4283). The recommended order ran lot-questions before the service designs, but its Lot 2a/2b and Lot 3 award criteria repeat each service's SDD and security answers, and the Lot 1a/1b quality answers draw on them. The order is now supplier-profile → social-value → service-design → sdd-lot* → pricing and security → lot-questions → declaration → review → submission-pack, in the commands' next steps and handoffs, the uk-gcloud-submission recipe (LOTQ now depends on the SDD, pricing and security), the dependency matrix, the README and the guides. lot-questions lists each service with whether its SDD and security evidence exist, warns when one is missing, and marks any answer that rests on a design alone for checking against the SDD. lot-questions and declaration listed only the designs with a G-Cloud Lot line, so a G-Cloud 14 design (no such line, or one saying "Lot 2") was skipped without a word, and lot-questions asked the user to guess its new lot. Both now share one listing that flags such a design as NOT A G-CLOUD 15 LOT and points to /arckit:service-design. Tests keep the listing identical in both commands, run it on a sample project, and check every lot lookup reads the G-Cloud Lot line. pricing's Lot 3 lookup now reads only each project's latest design. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Ports G-Cloud Kit v0.8.2's previous-framework check (977b9bd). review checked only that each document existed and what the design's lot line said, so a security document or supplier profile left over from G-Cloud 14 passed (both of G-Cloud Kit's test repositories had one). G-Cloud Kit added plugin/scripts/previous-framework.py; the overlay ships no scripts, so review and submission-pack run the same rules as one inline awk check. It recognises such a document by its structure: a title or opening quote naming an earlier G-Cloud, a Framework row or agreement number from one, a three-lot Lot value or a "1.3 Target Lot" checkbox, an SFIA rate card, the old minimum/maximum price fields, a supplier profile with no Central Digital Platform or PPON section, or a Public Contracts Regulations declaration. A G-Cloud 15 document that only mentions what changed isn't flagged. review blocks on every document it reports, submission-pack warns, and social-value, lot-questions and declaration warn when the profile they build on has no CDP or PPON section. Tests keep the two copies identical, flag sample G-Cloud 14 documents and pass every overlay template. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
… Lot 3 criteria Ports G-Cloud Kit v0.8.2's disqualification fix (e51ce2b). security said a tender is disqualified only when it marks under 33 on all four award criteria, while lot-questions, its template and guide said a zero on any one disqualifies. Attachment 2d v4.0 and Attachment 2 v5.0 both state the all-four rule for these lots (Attachment 2's general quality-threshold paragraph says any zero), and the listings fit it: 1,159 live Lot 2b services from 207 suppliers answer "Data sanitisation process: No". Every file now says a single zero loses that criterion's 2.5% and all four under 33 disqualifies. Lots 1a/1b are unchanged: a zero on either quality question disqualifies, as the question export says. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
…w does Ports G-Cloud Kit v0.8.2's review cross-checks (631a7b2). pricing says review checks its document against the SDD, but review only compared the pricing model, features and support levels. It now checks the education discount, the free trial answer, description and link, the Lot 1a/1b deployment models priced, that the Lot 2a/2b SDD holds no prices, the Lot 3 role levels against the supplier's rate card (ARC-000-RATE), and the lot-wide figures across the supplier's services in the lot. The review report gains a table for them, and each mismatch names both ARC- IDs and which one to change. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
…tch them to the checkboxes Ports G-Cloud Kit v0.8.2's social value fix (c4c4645, 357ee1c). social-value, its template and guide said the measures were worded exactly as the declaration's checkboxes. The model is built from the live listings, and comparing it with GCA's question export shows 17 of its 76 measures differ: the export's checkbox adds illustrative examples or a trailing clause, and one Outcome 6 checkbox adds a second measure. The overlay now records the listing wording (what buyers see), tells the user to tick the checkbox with the same opening words, notes the three checkboxes that share one listing text under Outcomes 6, 7 and 8, and never offers the three export fragments under Outcome 4. social-value now reads Section B of the declaration reference too, since it holds the checkboxes' full wording. social-value-model.md is G-Cloud Kit's regenerated copy, whose header no longer makes the old claim; the file is otherwise unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Ports G-Cloud Kit v0.8.2's option wording fix (3af61aa). The SDD templates said their option wording, taken from GCA's question export, "is what buyers filter on", but live listings word some options differently. Checked against all 42,893 listings, the SDD, service design and security templates now use the listing wording, with the Digital Platform's wording in a comment so the supplier ticks the right option: "Not a reseller", "Reseller providing extra features and support", "Reseller providing extra support", "Reseller (no extras)"; "Conforms to BS7858:2019" and "Other security clearance"; "Developed Vetting (DV)", "Security Clearance (SC)" and "Baseline Personnel Security Standard (BPSS)" without "Up to"; "API access"; "TLS (version 1.2 or above)" and "Legacy SSL and TLS (under version 1.2)"; "Complies with a recognised standard (for example CSA CCM version 4.0)"; "CSA CSM version 4.0" (GCA's spelling); "Google Apps", "Limited access network (for example PSN)" and "Dedicated link (for example VPN)"; "IOS" and "MacOS". The SDD commands, security, service-design, the framework reference, guides and SDD quality checks say the same, and a test stops the export-only wording coming back. Lot question options keep the export's wording: listings don't show them. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Ports the smaller fixes in G-Cloud Kit v0.8.2 (94641d3). - A stray `**` after the Lot 2a/2b data sanitisation question in the lot questions template. - compliance-frameworks.md said SC is required for OFFICIAL-SENSITIVE data. OFFICIAL-SENSITIVE is a handling caveat within OFFICIAL, which BPSS covers; SC is for SECRET and DV for TOP SECRET. The cloud-security skill's clearance table made the same claim and now says the same. - The Lot 3 pricing document is optional: 71% of the 27,496 live Lot 3 listings have one, while every Lot 1a, 2a and 2b listing does. The review and pricing templates, the Lot 3 SDD template and submission-pack no longer require it on Lot 3. - submission-pack wrote the supplier-level answers into every service's export. They are now written once per bid to projects/000-global/supplier/submission/bid-answers.md, refreshed only when a supplier-level document changes, and each service's answers-export.md points to it. - Every live listing shows a contact name and email (and all but 21 a phone number), the same on every service for 2,690 of 2,693 suppliers, but no template recorded it. The supplier profile now has a Listing Contact section, the SDD templates show it in G-Cloud Details (Document Control is ArcKit's shared partial), and review checks it. G-Cloud Kit's dependency-matrix change (pricing and security required for review) has no counterpart: the overlay's review already reports either missing as blocking. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
…'s Unreleased entries The Unreleased entries for the G-Cloud 15 port said the overlay carried G-Cloud Kit's rewrite as of v0.8.0. They now say v0.8.2 and describe what the v0.8.2 fixes changed: one category group per service, word limits on every free-text answer, one Lot 3 rate card per supplier (the new RATE doc-type), Cyber Essentials as a call-off condition, options worded as the live listings show them, the placeholder and previous-framework checks, the SDD-to-pricing cross-checks, the disqualification rule, social value measures in the listing wording, and the workflow order with the lot questions after the SDDs. The dependency matrix's recipe line says the same. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Regenerated with scripts/sync-claude-plugin-layout.py: the nested plugins/arckit-claude/plugins/uk/gcloud overlay, and the quality checklist and RENDERING.md partial every mirrored overlay carries (the RATE doc-type and the G-Cloud checklist changes). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn
Owner
Author
This push: G-Cloud Kit v0.8.2 fixes ported into the overlayThis PR ported the overlay from G-Cloud Kit v0.8.1. A test run on two real supplier projects (a Lot 2b SaaS and a Lot 3 support service) then found 15 defects, which G-Cloud Kit v0.8.2 fixes. The overlay had the same defects. This push brings the v0.8.2 fixes across, one commit per fix, adapted to the overlay's conventions: the supplier-wide LOTQ document, the What changed
Adaptations
Tests
🤖 Generated with Claude Code |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The
arckit-uk-gcloudoverlay was a port of G-Cloud Kit's G-Cloud 14 commands. G-Cloud 15 (RM1557.15) went live on 6 August 2026 under the Procurement Act 2023, run by the Government Commercial Agency (GCA, formerly CCS), and the Digital Marketplace now lists only G-Cloud 15 services. This PR ports G-Cloud Kit's G-Cloud 15 rewrite (gcloud-kit v0.8.0) into the overlay, adapted to ArcKit's projects, document IDs, Document Control, citations and quality checks. G-Cloud 14 is dropped.The overlay goes from 11 commands to 15:
sdd-lot1sdd-lot1a,sdd-lot1b(shared template)sdd-lot2sdd-lot2a,sdd-lot2b(shared template)sdd-lot3(SFIA rate card)sdd-lot3(DDaT rate card)social-value(new,SOCV)lot-questions(new,LOTQ)sfia-skillsskillddat-rate-cardskillWhat changed
Overlay (
plugins/arckit-uk-gcloud/)skills/gcloud-framework: G-Cloud 15 overview plusreferences/g-cloud-15/(per-lot service questions, lot questions, declaration, social value model, category trees), generated from GCA's question export.skills/ddat-rate-card(wassfia-skills): GCA's Lot 3 rate card (9 families, 58 roles, 222 levels), pricing rules, market rates for common levels. SFIA is kept as a supporting reference only.skills/cloud-security: G-Cloud 15 certifications by lot, new questions (post-quantum cryptography, Software Security Code of Practice, AI chatbot, FOCUS), current NCSC principle names.service-design: five-lot picker with category roots, records**G-Cloud Lot**: Lot <code> — <name>, handoffs to each SDD command by lot.sdd-lot1a/1b/2a/2b/3and their templates: rebuilt from GCA's export, section by section, with GCA's option wording.social-value,lot-questions(new): supplier-wide inprojects/000-global/supplier/. Lot questions are oneARC-000-LOTQdocument with a Part per lot group bid for.declaration: Procurement Act 2023 (CDP, PPON, exclusions declared on the CDP, debarment list, supply chain payments, modern slavery, bid-writer question shown but never answered for the supplier).pricing: 1a/1b price formula and minimum discount, 2a/2b discount bands, Lot 3 DDaT rate card and average day rate, forbidden-pricing check.security,review,submission-pack,supplier-profile,gcloud-competitors: G-Cloud 15 lots, limits and documents.uk-gcloud-submission: adds SOCV and LOTQ, defaults tosdd-lot2b.plugin.jsondescription and keywords.Core and site
doc-types.mjs,pages.md,_partials/RENDERING.md:SOCVandLOTQregistered (UK, Procurement, HIGH).references/quality-checklist.md: G-Cloud sections rewritten for G-Cloud 15; SOCV and LOTQ added. Synced into every plugin.guide-groups.mjs,guides.html,commands.html,index.html,README.md,docs/README.md,llms.txt,manifest.json,DEPENDENCY-MATRIX.md, both marketplace manifests, both changelogs (Unreleased).Deliberate differences from G-Cloud Kit
gcloud15-rate-benchmarks/gcloud15-discount-benchmarksCSVs are not ported. Lot 3 rates are compared with the market table in theddat-rate-cardskill (12 common role levels) or rival listings from/arckit:gcloud-competitors; Lot 2a/2b discounts have no benchmark.marketplaceMCP. WebSearch/WebFetch only, as before.lot-questions-{1,2,3}.md.**G-Cloud Lot**header line rather than a Document Control row (Document Control is the shared partial).gcloud-session.sh): the overlay has no hooks, and core ArcKit's health/navigator cover it.Tests
lint-markdown.ymlchecks pass (references, recipes, doc-type collisions and registry, shared-asset and guide parity, customize table, site links, llms.txt, quality-checklist refs and ladders, citation sections, doc-control resolution, colon style, agent frontmatter, orchestrator leaks, action pins, overlay layout).node --test tests/plugin/*.test.mjs tests/plugin/test_*.mjs: 537 passed; handoff validators andscripts/tests/*.mjspass.python scripts/converter.pythenpytest -q: 2287 passed, 228 skipped (pre-existing core-command skips).claude plugin validate --stricton both marketplaces and the overlay manifest, andclaude plugin validateon the overlay's commands and skills: passed.claude --plugin-dir) or the eval suite.🤖 Generated with Claude Code
https://claude.ai/code/session_019KL4LcSPGqEoVem8LZpTjn