[devin-ai-integration]

What and why

Audit of everything under examples/ against origin/develop at 76c98694c (40 PRs past v0.11.0), with binaries built from that commit by make build. Every tracked model (44 outside examples/self-model/) was run through -validate and -strict; every fenced ./bin/sysml, sysml, python and REPL block in the 20 walkthrough pages was replayed per block and cumulatively and diffed against the transcript that follows it; the inline sysml/kerml snippets were validated standalone, wrapped, cumulatively and as a part-def body. Each mismatch was re-run on the v0.11.0 release binary before being classed as documentation drift. Docs-only edits that keep each page's voice and elisions:

• Results in their effective multiplicity. A calculation or analysis result prints as a one-element sequence since 0.11.0 (fuelUsed = [120.0], = [6432.955324716369 [SI::'m/s']], result = [VerdictKind::pass], selectedAlternative = [Landers::relay (object #5)], -json "value": "[240.0]", -trace -> (130.0), sweep and explore tables). Re-recorded in analysis-demo, analysis-results-demo, runtime-showcase, disposal-robot-demo, relay-probe-demo, EXPRESSIONS-DEMO.
• Scheduling output. 1 choice point (1 independent), 592 choice points (591 independent), the (order-dependent; 1 choice point — explore shows the set) marker on sawBraking, and the check engine's 14748 states, 35626 moves (the page said 14664/35546; both binaries print 14748).
• Rendering. The Mermaid frontmatter now carries themeCSS/themeVariables with subGraphTitleMargin.bottom: 24, container titles are written on one line, the pins notice is gone, interconnection edges are unlabelled, -render-all byte counts moved, and a text state rendering names state def Modes unqualified; the filtered safetyView tree lists part def Thruster / part def Parachute without members (VIEWS-DEMO, disposal-robot-demo).
• REPL transcripts. %features of a completed state machine lists it under Behaviors: instead of <unknown> state entries, %advance 5 processes 2 events, %check of an unsatisfiable requirement prints its own verdict line before %explain, the %eval headings and the cumulative instance IDs of EXPRESSIONS-DEMO, and three lines -requirement … -satisfy no longer prints in disposal-team-demo.
• Python demos. lander_demo.py and robot_demo.py crashed on v0.11.0 and develop alike (TypeError: unsupported format string passed to list.__format__): the opensysml 0.11.0 client returns the one-element sequences the service now sends. They index the single element and their transcripts carry the — observed by run / — witnessed by run suffixes. This is the one non-Markdown change; the models are untouched, so the pilot differential baseline does not move.
• Index and figures. examples/README.md gains the general-views-demo/ row it lacked and names gridview-relationship-matrix.sysml; general-views-demo/README.md names the name-conflict warning vehicle.sysml:102 prints on every command; PARSER_FEATURES_DEMOS.md no longer restates the parser gate's count (it said 96/96, "94 + two extensions"; the gate is at 110/110) and links the roadmap's gate table instead, per CONTRIBUTING's "measured numbers have one home", and drops the development-session and task labels CONTRIBUTING forbids in reader-facing text.

Out of scope and untouched: examples/self-model/ and its tests.

Product gaps found (not papered over)

1. %instantiate mission / %features mission in relay-probe-demo prints subitems: <error: subsetting feature subparts of subitems: occurrence lifetime does not admit the change: cruise #3 would hold cruise #3, a whole it is a portion of, as a portion of itself> for the timeslice (both binaries); the page elides that region with ....
2. The Python client formats one-element results as lists (fuel left [130.0] kg): the scripts now unwrap, but a scalar-valued out parameter arriving as a list is worth a client-side decision.

A stale-records regression first listed here (-record-run emptying the Runs that no longer match the model table) did not reproduce on a clean make build of 76c98694c — the renders with and without -record-run both list relayRun; the earlier observation came from a -dirty working-tree binary and is withdrawn.

Overlap

Open PRs touching examples/: #1502 (docs/self-model-surfaces-refresh @ 304df41bb), #1484 (fix/part-usage-part-definition-reading @ 70dae7174), #1477 (feat/document-whole-bodies @ 2672d357c) — all only under examples/self-model/; each dry-merges cleanly against this head, as does origin/develop at 76c98694c.

How it was verified

At a5892bc02:

• python3 scripts/check-doc-links.py → 0 broken link(s); python3 scripts/check-doc-ids.py → No internal work-item labels…; python3 scripts/check-doc-figures.py → ok; python3 scripts/changelog.py check → rc 0
• make docs-landing-assets → rc 0; make docs-check → 0 broken link(s), rc 0; python -m mkdocs build --strict → rc 0
• go test -count=1 ./examples/ ./tests/hygiene/ → ok, ok
• go test -run TestStdlibConformance ./internal/workspace/libs → 110/110 clean
• Every edited transcript re-replayed with the harness against ./bin/sysml at 76c98694c; lander_demo.py output is byte-identical to its README block, robot_demo.py differs only in the elided condition text; both run against bin/sysml-grpc via OPENSYSML_BINARY.

Checklist

• make test and make lint pass locally — docs and example-script change; the gates above ran
• Tests added or updated for the change — none needed (no code under test changed)
• Documentation extended where it already covers the surface
• Changelog entry added as changes/unreleased/docs-audit-examples.fixed.md
• baselines regenerated and make docs-counts run if a gate count moved — no gate count moved
• No internal work-item labels in the body, docs, or changelog

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring