Navigation menu icons: correct the stale "not scriptable" claim, and make row 8 judge the menu - #125
Merged
Merged
Conversation
…a full Stage-5 build (second occurrence) Process finding from a 3-module requirements-driven build: gate-check never re-run after scaffold, UI skills reachable only through undispatched role-agent stubs, instruments never installed, and the 2026-09-09 inbox note predicting this was still unpromoted. Five enforcement proposals, smallest first. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XgoKZ42wYcMCBPG8Gj2PCh
…ver executed Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XgoKZ42wYcMCBPG8Gj2PCh
… a LOOK pass conversion-runbook.md §1b rule 8: every stage and every module build ends on bin/gate-check.sh, run and pasted, with zero PENDING or each PENDING named and waived. The obligation check already reports a pass nobody performed — but only on a run, and nothing before this required one. ui-preflight-pages.md: page-fidelity.js matches identifiers in page MDL against wireframe text and cannot see placement. On the same commit, two screens scored 100% fidelity against 58% and 40% on an element checklist; a third ran the other way. Use the score for the rework curve between LOOK passes, never in place of one. init-project.sh: the generated CLAUDE.local.md now carries a before/after page-MDL item, so the pre-flight text reaches the agent that writes pages. Field evidence: lowcode-vs-highcode-benchmark results/mendix-run1 (closed blind), results/mendix-run1b (seven screens at 0/18/40/40/58/58/69% baseline), and results/highcode-run1b as the non-Mendix control (43% and 56% baseline on two screens) — a pipeline gap, not a platform one. Withdraws proposal D of the 2026-09-19 inbox note as field-refuted. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XgoKZ42wYcMCBPG8Gj2PCh
…d in e466e4c Fidelity scored 100% on the two screens an element checklist put at 58% and 40% on the same commit, so a fidelity threshold cannot discharge the look obligation. Cites results/mendix-run1b and results/highcode-run1b. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XgoKZ42wYcMCBPG8Gj2PCh
…ia ICON The skill asserted that navigation.create has no menu-item icon token and that per-item icons must be assigned by hand in Studio Pro. mxcli v0.22.0's `syntax navigation.create` documents the ICON clause, the quoting rule for hyphenated Atlas names, and SHOW/DESCRIBE ICON COLLECTION. Field cost: the PRD benchmark's Arm A shipped five text-only nav items through a full build and two UI sweeps because the skill said icons were not scriptable, so nobody re-probed the binary. Corrected in place, with the probe command and a worked example, and flagged as the capability-probe rule failing where it hurts most. Also separated the MDL half (ICON places the glyph) from the CSS half (icon-only-when-collapsed), and told readers to take Atlas sidebar class names from their own themesource, not another project's theme. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01XgoKZ42wYcMCBPG8Gj2PCh
…v bar Rubric row 8 already caught an un-skinned default nav bar and a sidebar design shipped as a top-bar app. It said nothing about the navigation profile's own item list. That list is chrome on every screen and belongs to no page, so a page-by-page LOOK pass structurally never reaches it. Field evidence: Arm A of the PRD low-code-vs-high-code benchmark passed a full module review and two UI sweeps carrying five bare text menu labels with no icons; the user's verdict on the shipped nav was "lots of text ugly stuff". The icon capability existed the whole time. Row 8 now names labels, icons, order and grouping as a finding class, instructs judging the menu once explicitly on the first screen opened, and points at learned-sidebar-collapse-icons.md so the corrected icon capability is reachable from the review rubric, not only from a build skill. CHANGELOG line in the same commit.
…e-darwin-i3gfue # Conflicts: # CHANGELOG.md
…e-darwin-i3gfue # Conflicts: # CHANGELOG.md
… from the PRD benchmark
# Conflicts: # CHANGELOG.md
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
2 participants
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.
Two fixes from a field build where a full module review and two UI sweeps all passed while the app shipped five bare text labels and no navigation icons.
What changed
skills/learned-sidebar-collapse-icons.md— the file claimed menu-item icons were not scriptable and had to be set in Studio Pro. That is false on mxcli ≥ v0.22.0, and the claim had been carried forward from an older binary. Corrected with the verified grammar and worked examples:MENU ITEM 'Label' PAGE Module.Page ICON Module.IconCollection.NameICONtakes a model reference, not a string; hyphenated Atlas names need double quotes (ICON Atlas_Core.Atlas."check-circle")SHOW ICON COLLECTION/DESCRIBE ICON COLLECTIONthemesource/atlas_core/, never inside the generatedmxcli:theme:beginblockThe correction is written as a probe instruction rather than a new assertion, per the capability-probe rule: a general prior is not evidence about this binary.
skills/module-review.md— rubric row 8 judged the navigation bar (top vs side) but never the navigation menu (labels, icons, order, grouping). The menu is chrome on every screen and belongs to no page, so a page-by-page LOOK pass structurally never reaches it. Row 8 now says to judge it once, explicitly, on the first screen opened, and points at the corrected icon skill.Why it matters
This is a structural hole, not a discipline failure. The rest of the guidance was already present and correct —
ui-preflight-pages.mdhas a shell check, and row 8 already named the top-bar/side-bar mismatch. Nothing anywhere said to look at the menu's own items, so nothing did.Changelog
Both entries are in
CHANGELOG.mdunder## Unreleased, credited to the PRD benchmark, Arm A (Maurits Visser).Merge
masterhas been merged into this branch twice to resolveCHANGELOG.mdconflicts as the release cycle and new entries landed upstream — most recentlyccd2ed1(againstmasterat598e38a). Both resolutions keep both sides verbatim: this branch's two entries sit at the top of## Unreleased, followed by every linemasteradded. Nothing from either side was dropped.State at this head: mergeable, no conflicts; CI
checksgreen;bin/render-routing.sh --checkpasses on the merged tree (routing surfaces in sync, all skills routed or exempted, baseline within budget).