SkillAgentSearch skills...

scenario-zbrush-automation

Use when automating ZBrush 2026: batch processing a folder of OBJ or ZTL files, a ZBrush Python script or ZScript, recording a macro to learn a palette path, the ZBrush -script command line, scripting UV Master, Decimation Master or Multi Map Exporter, exporting every SubTool, GoZ or GoB round trips…

Install / Use

npx skills add scenario-labs/skills --skill scenario-zbrush-automation

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

86/100

Category

Automation

Supported Platforms

Universal

Our assessment of scenario-zbrush-automation

scenario-zbrush-automation scores 86/100 on our quality scale, 1791st of 2,758 Automation skills we index.

Its SKILL.md is 19 KB long, well organised into 10 sections and no code examples: a thorough specification that gives an agent plenty to work with.

With 1,048 GitHub stars, it is one of the more widely adopted skills in the catalogue.

Substance
30/30
Structure
13/20
Description
15/15
Adoption
13/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated 35 days ago, so scenario-zbrush-automation is actively maintained.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands.

Automated pattern scan on 2026-11-20. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

scenario-zbrush-automation compared with similar skills

All 4 of these similar skills score higher than scenario-zbrush-automation; compare them before choosing.

SkillScoreStarsUpdatedFormat
scenario-zbrush-automation (this skill)by scenario-labs861.0k35d agoSKILL.md
Agent-Reachby Panniantong100114.0ktodayCLAUDE.md
rufloby ruvnet10082.3ktodayCLAUDE.md
headroomby headroomlabs-ai10080.5ktodayCLAUDE.md
Claude-Code-Game-Studiosby Donchitos10040.6k12d agoCLAUDE.md

Frequently asked questions

How do I install scenario-zbrush-automation?
Run npx skills add scenario-labs/skills --skill scenario-zbrush-automation. The install tabs above show the steps for each supported agent.
Which AI agents does scenario-zbrush-automation work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is scenario-zbrush-automation safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is scenario-zbrush-automation still maintained?
The repository was last updated 35 days ago, so scenario-zbrush-automation is actively maintained.

name: scenario-zbrush-automation description: "Use when automating ZBrush 2026: batch processing a folder of OBJ or ZTL files, a ZBrush Python script or ZScript, recording a macro to learn a palette path, the ZBrush -script command line, scripting UV Master, Decimation Master or Multi Map Exporter, exporting every SubTool, GoZ or GoB round trips with Maya or Blender, renaming SubTools, or batch reports. Also when a plugin press never returns, a dialog stalls a batch, one file kills the whole run, or GoZ updates the wrong SubTool." license: MIT

ZBrush automation (tools and pipeline TD)

An expert ZBrush TD automates with contracts and proof: every file is its own tool, every step is a separate call with a time limit, every output is read back from disk, and a hung plugin costs one file, never the batch. ZBrush has no timeout, no undo control from Python and dialogs that code cannot answer, so the safety lives on the agent side: the runner, the checkpoints, the report. If a sibling skill named here is missing from your available skills, load it with skill_get (Scenario MCP) or ask the user to install it (https://github.com/scenario-labs/skills#install); unattended, proceed from tool schemas and flag the gap.

REQUIRED BACKGROUND: scenario-zbrush-expert (bridge, zb_launch, zb_ops, zb_audit, review loop, 2026 traps).

Status (2026-09-24, refactor pass 2): every ZBrush call here is not yet run in ZBrush (no session assigned; the README's startup stall is still open). Offline: 56 tests pass (tests/code/zbrush-automation/run_offline.py), including a GoZ reader and writer matching ZBrush's own ZResources/DXFStar.GoZ byte for byte. Live suite: tests/code/zbrush-automation/run_live.sh.

Stance (the expert delta)

  • One file, one tool, one step per call. "Only one zscript/plugin can be active at a time, so if your zscript calls someting like UV Mapper control passes to UV Mapper and no more commands will run" (marcus_civis, ZBrushCentral 2014); pressing Decimation Master's Pre-process ended ijacobs' script (2013) and forced cgside into a two-run helper (2022, [00:06:56]). So the batch is an agent-side loop (zb_batch.run_folder) that sends each step as its own bridge call with its own timeout, pings after a timeout, and restarts ZBrush after a hang. One long in-ZBrush script dies with its first stuck plugin. [added design]
  • Plugins last, checkpoint first. cgside saves before ProjectAll "since it might crash the program" [00:02:19]. zb_ops.project_all refuses to run without a versioned save and the recipe's project step passes one ({stem}_preproject.ztl), even with save_ztl off. Decimation Master runs last, after every valuable output is on disk, so a hang there leaves a partial file, not a lost one. [added]
  • Leave the interpreter as found. ZBrush has one Python for the session and the script folder is not on sys.path (Maxon Style Guide). Send code only through remote_call/zb_launch.run: modules are imported by path, then sys.path, sys.modules, stdout and the working directory are restored, also after an error. End a batch with zb_launch.hygiene() (empty lists).
  • Dialogs are the automation killer; budget them before running. A studio dropped ZBrush from its farm over Note dialogs (danvas 2017); "no way to detect whether they succeed or error" (BigRoyNL 2024); "if the export plugin, trigger a popup you are screwed" (m_adam, Maxon 2025). set_next_filename feeds built-in buttons only: FBX ExportImport shows options "You cannot escape" (ferdinand, Maxon 2025) and USD Format ignores the preset (usd-portal). Unattended exports are OBJ or .GoZ by path; has_next_filename() still true after a press means a dialog is probably waiting [added].
  • Names are identity. GoZ needs unique names "including between all the loaded Tools" (GoZ docs); Decimation Master caches and UV Master control maps are keyed by the tool name, so a new tool with an old name inherits stale data (plugin docs). Stage each input under a unique, sanitized name before import (stage_input); on this Mac's case-insensitive disk rock.obj and Rock.OBJ are one file [added].
  • Learn paths from ZBrush, never guess. Recordings show exact paths and hidden preconditions (Maxon ZScript manual; MadPony DnLlDJxfxHs [00:02:16]); Ctrl+hover shows a path. For an agent the cheapest oracle is ZBrush's own Activity log, which writes every press and set with its canonical path (zb_batch.ActivityWatch; proven by the 05:20 log). Then zb_plugin_ops.probe() confirms exists, range and title.
  • Record once, apply to all. Pavlovich replays one recording on every visible SubTool with ZRepeat It [00:37:30]; ZRepeat It is not shipped in 2026.2.1, the Python equivalent is for_each_subtool.
  • Files, not live links, between apps. GoZ sends the lowest subdivision level (GoZ docs; FlippedNormals [00:04:02]), updates by name, and raises a reprojection prompt when topology changed. An agent hands over .GoZ or OBJ files plus a manifest and lets the other app's agent import them.

Establish first

| Input | Changes | Default when silent | | ----------------------------------------------------- | ---------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Inputs: count, format, faces per file | timeouts, DynaMesh resolution, restart cadence | OBJ, audited first with zb_audit | | Outputs: low count, map size and space, preview count | recipe params | brief's numbers; else DEFAULT_PARAMS [added] | | Parts per file, meaningful polygroups | ZRemesher grouping | cgside: Auto Groups then ZRemesher Keep Groups on the copy, so the low follows the parts (MyhwQvkcnwI [00:08:38]); zr_auto_groups=False when the input's groups matter | | Target app: Maya, Blender, Unreal, Unity | flip V, green channel, axes, format | Maya: OBJ + tangent PNG flipped V, OpenGL green [added] | | Whose ZBrush | restart rights | a ZBrush the batch did not start is never restarted or quit | | Watched or unattended | pilot size, screenshot use | pilot of one file, watched | | Shared config allowed (GoZ folder, Macro palette) | GoZ push, key-answer macros | no: ask first |

Workflow

  1. Session and probe. zb_launch.start() or reuse a running bridge; zb_plugin_ops.probe() on every path the recipe uses; memory(). GATE: ping returns 2026.2, every recipe path exists, plugins present.
  2. Plan dry. zb_batch.plan(files, out, params) resolves every file's steps, paths and timeouts without ZBrush. GATE: outputs outside the install (read-only since 2026.1), unique stems, no FBX or USD step.
  3. Pilot one file, watched. run_folder([f], out, params); open sheet.png and the normal map; import the low and map in the target app (scenario-maya-pipeline-scripting or a Blender skill). GATE: file ok, step timings recorded, map orientation confirmed in the target app (flip V and green are [verify] until then).
  4. Run the batch. Bridge mode by default; mode="script" (one ZBrush -script process per file) once live_a05 has shown how ZBrush exits. Drop a STOP file in the output folder to end between files; rerun to resume (sha1 and recipe hash). GATE: report.json written after every file.
  5. Recover. For each failed or partial: read steps[].outcome (error, timeout_recovered, hang), the hang evidence (screenshot when the screen is unlocked; ZBrush Notes are drawn in the canvas, only the screenshot shows them), and batch.blocked. Fix params or paths, rerun: finished files are skipped. GATE: every non-ok file has a cause in the report.
  6. Review and deliver. Open the contact sheet; check low_audit, preview_audit, normal_stats; write the handoff manifest. GATE: numbers pass, sheet looked at, report and manifest delivered.

Numbers

| Value | Relative to | Source | | -------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------- | --------------------------------------------- | | UV Master: 100k to 150k faces max; seconds at 3k, up to 5 min at 150k | faces of the SubTool unwrapped | UV Master doc | | Divide multiplies faces by 4; levels = smallest n with low x 4^n >= source faces, top capped at 8M | projection stack | Pablo Logic Part 7; cap [added] | | Decimation % of decimation is kept percent of points; 100 none, 0.01 max | points; about 2 triangles per point, so % = 100 x target_faces / 2 / points | DM doc; ijacobs 2013; Euler [added] | | DM quality: 40 to 100 % looks the same, 2 % still good, 1 % degrades | original model | DM doc | | ZRemesher target in thousands, approximate (5 gave 4,963) | target polygons | Pablo Logic Part 7 | | Project All Dist 0.1, set explicitly | same shape, new topology | Drust nxMYYsyJt3o [00:02:19]; toolkit default | | create_normal_map(..., local_coordinates=True) = tangent; default False = world | map space | SDK stub | | Macro file name 8+ characters, in a subfolder; new folder needs a restart | <Asset Dir>/ZStartup/Macros/ | Maxon interfaces doc | | Step timeouts 30 to 1,800 s, raised to 4 x slowest observed, 10 x base max; restart every 10 files | per step, per session | [added] | | Tangent map sanity: blue >= 128 on 90 % of pixels | final PNG

Truncated for display — read the full file on GitHub.

Related Skills

View on GitHub
GitHub Stars1.0k
CategoryAutomation
Updated1mo ago
Forks129

Languages

Python

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions
scenario-zbrush-automation — Universal Skill: Install & Safety Check | SkillAgent