skills: Mendix best-practices index, microflow tiers and split-first, mxcli layout finding

[MendixMau]

What changed and why (one paragraph)

Follow-up to #135. skills/mendix-best-practices-index.md (new) answers "is there a Mendix best practice for this, and what enforces it here?" with one row per area: the Mendix docs page that states the practice, the section of the mxcli-bundled assess-quality skill that summarises it, the toolkit skill that applies it before the write, and the mxcli lint rule that catches it after exec. It is an index, not a copy (CLAUDE.md authoring rule 5): 17 URLs verified HTTP 200 on 2026-09-25, Mendix's own Best Practice Recommender rules (MXP001–016) anchor the performance rows, and four rows say out loud that no lint rule exists and the preflight checklist is the only check. skills/microflow-preflight.md now decides a tier before any MDL — Simple / Guided / Split-first — and Split-first posts an orchestrator + one SUB_ per responsibility, with signatures, that the user confirms before the first MDL line (prompted by a colleague's session refusing a long microflow as "too difficult" while the same task went through on a stronger model: the piece was too big, not the task). agents/mdl-agent.md gains two rules: never hand back "too difficult", hand back the split plan; escalate one failing SUB_ by name after one retry, never the whole script. The mxcli team's answer on positioning — a standalone mxcli layout command — is recorded in the preflight skill, learned-microflow-patterns.md and the if-branch bug draft: on v0.24.0 and upstream main at 2026-09-25 that command arranges domain models only, so the no-@position rule and the workaround stand until a microflow mode ships. Routing row (all agents, stages 3/5/6, group reference) plus the re-rendered surfaces.

Field evidence

• mxcli layout --help on v0.24.0 (built from the tag) and on upstream main (d3adfa4, 2026-09-25): "Arrange the entities of a domain model" — no microflow/nanoflow mode. mxcli layout -p app.mpr --dry-run on a scratch copy of a PoC model (Mendix 11.12.1): 10 entity moves, no microflow.
• mxcli lint --list-rules on the same binary is the source of every rule ID in the index.
• curl -sI on every URL in the index: 17 × HTTP 200 (redirect targets used, e.g. extracting-and-using-sub-microflows, qsm).
• bin/render-routing.sh --check exit 0; check-pr-discipline, check-scripts, check-portability, leak guard: clean.

Checklist

• No client data — leak guard passes; a manual grep of the diff for client, person and home-path patterns is empty.
• Size: 16 files / 66 lines. Eight of the files are render-routing.sh outputs from the one new routing row; the hand-written change is 8 files.
• Test tier: T0 (inspection) — no script under bin/ changed except the rendered gate-check.sh stage map; the guide-reopen regex count on the four guarded files is unchanged vs master (4).
• Instrument rules: n/a (no new instrument).
• Routing row for the new skill in bin/lib/skill-routing.tsv, surfaces re-rendered.
• CHANGELOG lines in the same commit, credited.

🤖 Generated with Claude Code

https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw

---

Generated by Claude Code