scenario-unity-pipeline-automation

Unity pipeline automation (tools and build engineer)

Expert level here means a pipeline that gives the same result on every machine and every run: rules applied at import time, edits made through serialization, content packed by when it loads, tests that gate the build, and a build that CI runs from a clean checkout. The stance: nothing is done by hand twice, nothing is trusted without a count, a test or a frame, and a second run changes nothing. If a sibling skill named here is missing from your available skills, ask the user to install it (npx skills add scenario-labs/skills --skill <name>); unattended, proceed from tool schemas and flag the gap.

REQUIRED BACKGROUND: scenario-unity-expert (channels, job protocol, review loop, 6.3 traps). Import its toolkit; this skill adds scripts/ut_pipeline.py and scripts/AgentKit/Pipeline/.

Status (2026-09-24, after the blind-grade refactor): every procedure ran in Unity 6000.3.21f1 on this Mac against a 200-FBX Blender drop: tests/code/unity-pipeline-automation/ (results in archive/tests/unity-pipeline-automation/live_results.jsonl).

Stance (the expert delta)

1. Import rules are code that runs at import, and deterministic, and you prove it. Settings in OnPreprocess*, GetVersion() bumped on every code change, config declared before it is read (context.DependsOnSourceAsset), no clocks, no dictionary order, no asset writes in callbacks (6.3 Manual; Javier Abud Chavez, S2P9n5U9xVw [00:04:35], [00:19:29]). Proof: -consistencyCheck before trusting a shared cache or CI (Manual). Observed: 12,184 assets checked in 23.5 s; a random-pixel probe texture flagged, 0 of the 593 governed files. A bump or a new postprocessor reimports every asset of that importer type, packages included (1,513 textures).
2. An edit counts only if it goes through serialization, on the exact object. Record Undo on the component that changes, before the change (Freya Holmér, pZ45O2hg_30 [03:01:01]). Observed: Undo.RecordObject(go) then a move is not undone; RecordObject(go.transform) is. A direct write on a prefab instance reverts at the next prefab update; a ScriptableObject is not saved without SetDirty.
3. One AssetDatabase batch per stage, measured. StartAssetEditing / StopAssetEditing in try/finally (Abud Chavez [00:30:21]), and SaveAndReimport / ImportAsset inside it too. Count with the AssetDatabase counters, not the log (Abud Chavez [00:26:44]). Observed: two remap passes over 195 FBX, 19.0 s one by one vs 5.1 s batched.
4. Pack Addressables by loading condition, not by folder, in a stable order. Assets needed under exactly the same conditions share a group; dependencies used by several groups are placed explicitly (Olle Axelsson, Valheim, Fzu2Das_1FU [00:23:08]). Unassigned shared dependencies are copied into every bundle [00:18:44]; observed, that includes URP's Lit.shader. Assign in ordinal order (6.3 Manual, Addressables determinism): 2.9.1 sorts groups and entries by GUID when it saves, but the label table keeps insertion order.
5. Released is not unloaded, and references are loads. Memory returns when the bundle's refcount reaches zero (Addressables 2.9 manual); a build scene that references Addressable content loads it with the scene and again from the bundle (Code Monkey, C6i_JiRoIfk [00:03:32]; Axelsson [00:24:49]). Gate it; measure unloads in the player.
6. Fix the platform at launch, the same on every stage, and ship the content you built. Target switches inside a batch job do nothing (Manual; Thom Hopper, BlVsi2cSJ88 [00:27:18]); build target is an import dependency (Abud Chavez [00:15:49]). Observed: one stage on Android reimported 1,717 assets in 52 s; switching back reused cached artifacts (60 imports). Addressables' default settings rebuild all content inside every player build.
7. Tests are the gate. Content rules as parametric Edit Mode tests, one case per asset (Warnecke and Fine, wTiF2D0_vKA [00:24:28]), run synchronously in a pre-build callback that logs the pass [00:31:43]. Any logged error fails a test unless LogAssert.Expected (Manual); observed, that includes an import error logged by a reimport inside the test. Never -quit with -runTests.
8. Release builds come from a clean tree, without debug defines, with provenance kept (docs digest checklist; Hopper, profile defines). Narrow named tools over generic execution (git-amend, VPjo-M6mPkE [00:03:54]; Unity, DgNrgZeJOxQ [00:07:41]).

Establish first

• Drop contract: DCC and export settings, folder per category, suffixes, units. Default: Blender "FBX All" + "Apply Transform", _BaseColor/_Normal, a manifest.csv for extra labels.
• Loading conditions: loaded as a set (level, biome) or one by one? Default: labels, key_by="label"; streaming uses key_by="root".
• Targets and budgets: platforms (one -buildTarget per pipeline run), texture cap per platform, vertex budget, bundle budget per label. Defaults: 1024 px desktop, 512 mobile, 5,000 vertices.
• Content updates: remote content after release? Archive addressables_content_state.bin per platform per release.
• Channel and CI: is the user's editor open on the project (ut_live.channel(P))? Git repo? Accelerator? License: Personal (Hub sign-in only) or paid seat.

Workflow

1. Preflight. ut_env.preflight(P); ut_pipeline.install(P). Editor open: the same jobs through ut_live.call or the [CliCommand] tools. Pipeline code a team's tests must call: its own Editor asmdef plus a test asmdef (install_asmdef_template). GATE: Echo ok, zero compile errors (one aborts every batch job).
2. Scan the drop offline. up.scan_drop(drop), same rules JSON. GATE: every reject has a reason; duplicates reject every copy.
3. Import. up.import_drop(P, drop, target=T): changed files into <root>/<Category>/<Name>/, one batch, audit. GATE: accepted = found - rejected; audit errors explained; after a postprocessor change, up.consistency_check(P, target=T) flags nothing under Assets/.
4. Materials and prefabs. up.build_prefabs(P, target=T): URP Lit per prop, keywords by ShaderGUI.ValidateMaterial, FBX remaps with SaveAndReimport inside the batch, Prefab Variants built in a preview scene, saved only when a value changed. GATE: prefabs = accepted - size errors; a rerun rewrites 0 files (git status clean).
5. Look at it. EditorEditJobs.BuildLineup then AgentCapture.CaptureViews with graphics=True, ut_review.review_capture, open the sheet. GATE: no magenta, black or blank flags; props upright, grounded, in scale.
6. Addressables groups. up.assign_addressables(P, "condition", target=T). GATE: implicit_duplicates == 0; up.hard_references(P) ok.
7. Content build. up.build_content(P, T): builder index >= 0, ClearCachedData + BuildCache.PurgeCache(false), DoNotBuildWithPlayer, layout read back, state archived. GATE: duplicated assets 0, bytes per label within budget.
8. Tests. ut_run.run_tests(P, "EditMode", categories="PipelinePreBuild", extra_args=["-buildTarget", up.cli_target(T)]). GATE: failed 0, total equals the expected asset count.
9. Player build. Profile copied from a committed template (PipelineBuild.CreateBuildProfile with template), ut_run.build(P, T, profile=...), or raw -activeBuildProfile ... -executeMethod AgentKit.Pipeline.PipelineBuild.BuildFromCommandLine. Release: -agentRelease (gate refuses debug defines), clean tree, up.release_ready(repo, app). GATE: AGENT_GATE ok with hard_references 0, BuildReport Succeeded.
10. Smoke the player. up.player_smoke(app, label). GATE: loaded assets equal the manifest count; bundles unload after release.
11. CI. up.write_ci(repo): ci/pipeline.sh (one process per stage, the same -buildTarget on each; ACCELERATOR=, CONSISTENCY=1, RELEASE=1) and GitHub Actions files. GATE: passes from a checkout without Library/; license returned if: always(); with Accelerator, the log's cache server= count above 0.

Numbers

Value	Relative to	Source
5.6 to 6.7 s	593 files copied and imported in one batch	observed
19.0 s vs 5.1 s	two remap passes over 195 FBX (SaveAndReimport), one by one vs in one batch	observed
11 to 12 s	197 materials, 197 FBX remaps, 196 Prefab Variants, in-editor time	observed
1,717 imports, 52 s / 60 imports	first launch on another platform / switching back (cached artifacts)	observed
1,513 textures, 17 s	reimported by one GetVersion() bump of a do-nothing texture postprocessor	observed
12,184 checked, 23.5 s	-consistencyCheck of this project (7,390 skipped; 2 URP shaders flagged: baseline packages)	observed
0 files	changed by a second full run (import, prefabs, groups)	observed
791 tests, 10 to 16 s	content tests for 196 prefabs, 396 textures, 197 models	observed
13.5 / 12.9 / 4.8 MB	worst label load, category / isolate / condition layouts (3 biomes)	observed
5	duplicated assets in the category layout (2 textures, Lit.shader, 2 fallbacks)	observed
0 to 63 ms	bundles still loaded after the last Release (7 player runs)	observed
60 to 158 s / 5 to 23 s	first macOS Mono player build / incremental rebuilds (shared machine); 127.4 MB player	observed
0.4 to 1.2 s vs 10 to 17 s	the same job through a resident editor's bridge vs a batch job with editor start	observed
4 GB	ceiling per bundle for cross-platform compatibility	Addressables 2.9 manual
30,000 vertices	example FBX budget warning	Unity tutorial, qZ5NmyoxKRk [00:03:55]
38%, about 110 MB	Valheim memory cut from streaming, then from fewer loaded bundles	Axelsson [00:03:02], [00:30:41]
179 s / 486 s	generated ci/pipeline.sh from a checkout without Library/ (v0.2, quiet machine / v0.1, 6 to 8 other editors)	observed
0 / 8 / 6	unity test exit codes: pass / tests failed (never retry) / no verdict (retry)	unity-cli skill

Quality gates

• Measurable: envelope ok, no compile errors, active_target equal on every stage; ImportDrop counts; consistency check clean under Assets/; prefab count; rerun: copied 0, git status clean; implicit_duplicates 0 and duplicated_assets 0; per-label KB; no build scene hard-referencing Addressable content; NUnit failed 0 with the expected total; AGENT_GATE ok; BuildReport Succeeded; player smoke counts; release: clean tree, provenance source.dirty false, no forbidden defines; CI script exit 0 from a clean checkout.
• Visual: lineup sheet opened: not magenta, normal maps shading, props upright, pivots on the ground, sizes plausible next to the 1.8 m capsule.

Common mistakes

Mistake	What it looks like	Fix
importer settings in OnPostprocessTexture; code changed, GetVersion() not bumped	old settings used; machines disagree	OnPreprocess*; bump the version; -consistencyCheck
non-import keys added to the rules JSON	every governed asset reimports	separate file (pipeline_checks.json)
Undo.RecordObject(gameObject) before moving it	move not undoable	record the Transform
direct write on a prefab instance or a ScriptableObject	reverts at the next prefab update / file unchanged	RecordPrefabInstancePropertyModifications / SetDirty
StartAssetEditing without finally; SaveAndReimport per FBX outside a batch	frozen database; 4x slower remaps	try/finally; reimport inside the batch
stages without the same -buildTarget	each stage reimports textures and shaders	the same target on every stage, tests included
one group per folder with shared textures	the same texture and shader in every bundle	condition packing; build layout check
labels added in file or HashSet order	the settings file differs between machines	ordinal order
a build scene references an Addressable prefab	loaded twice, bundles bypassed	HardReferences in the gate
content built, then a default player build	content rebuilt, archived state stale	DoNotBuildWithPlayer + gate
SetActiveBuildProfile in a batch job; profile from the internal API	old target or defines; breaks on upgrade	-activeBuildProfile; copy a committed template
a test reimports an asset whose importer logs an error	"Unhandled log message", test fails	LogAssert.Expect
tests in a test asmdef calling code in Assembly-CSharp-Editor	CS0103, every batch job aborts	pipeline code in its own Editor asmdef
-quit with Accelerator, no wait flag; server unreachable	uploads lost; job passes silently	-cacheServerWaitForUploadCompletion; check cache server=
release from a dirty tree	provenance dirty: true, the CLI still builds	clean tree, RELEASE=1, gitignore Addressables' link.xml
Blender default FBX export	root rotated 90 degrees, scale 100	FBX All + Apply Transform (scenario-blender-export)
duplicate name resolved by order	the wrong asset silently wins	reject every copy, ask
Addressables settings created inside a job	the next run rewrites the settings file once	set the two provider types after GetSettings(true)
-quit with -runTests; relative --output-path	no XML; build outside the project	omit -quit; absolute path

Handoffs

• Receives: FBX and textures with export settings from scenario-blender-export, scenario-maya-expert, scenario-3d, scenario-textures; code layout and asmdefs from scenario-unity-architecture; which props load together from scenario-unity-world-building.
• Delivers: to scenario-unity-performance the build report, bundle sizes, per-label loads; to scenario-unity-mobile texture formats (this skill sets size caps only) and signed builds; to scenario-unity-web Web builds (no WaitForCompletion on WebGL); to scenario-unity-rendering-lighting lightmap UV needs; to the lead: rules JSON, props manifest, envelopes, NUnit totals, layout JSON, BuildReport, provenance, contact sheet.

Unity 6.3 notes

• Addressables 2.9.1 is this editor's default (folder keys and await handle are 4.0); its postprocessor adopts any imported group asset into the default settings. Content Directories replace AssetBundle builds in 6.6.
• IPreprocessBuildWithContext also fires for AssetBundle builds: filter with BuildPipeline.isBuildingPlayer.
• No public profile-creation API before 6.5; AssetDatabase counters are public in UnityEditor.Experimental.AssetDatabaseExperimental.counters.
• The Unity CLI is beta; unity build writes its provenance manifest inside the .app on macOS. Personal licenses activate only through Hub sign-in.

References

• references/procedures.md: the pipeline as copyable calls, each with its live test and result.
• references/expert-notes.md: principles and judgment by expert, with source and timestamp.
• references/critique.md: the rubric to judge a pipeline, a layout, a build, a release.
• references/gui-paths.md: windows and menus for the same tasks.
• references/sources.md: every source with credentials, URLs, timestamps, revision history.
• scripts/ut_pipeline.py, scripts/AgentKit/Pipeline/*.cs (jobs, postprocessor, checks, gate, content tests), scripts/AgentKit/PipelineCli/, scripts/Runtime/, scripts/templates/PipelineAsmdef/: the code.