Skip to main content

dcc-mcp-speedtree

Create and edit native SpeedTree generator graphs, regenerate them with the licensed Modeler command-line exporter, and hand verified assets to another DCC-MCP adapter. Graph authoring uses the observed SPM/STT file format and is experimental; full live Modeler control is not yet available.

Part of the DCC-MCP host matrix

dcc-mcp-speedtree — SpeedTree adapter for DCC-MCP — edit experimental native generator graphs, run the licensed Modeler CLI and verify vegetation exports.

It is one of 48 host adapters in the DCC-MCP catalog. Every adapter speaks the same MCP protocol and builds on the same core runtime contract; each one exposes the tools its own host needs on top of that.

This block is generated from the catalog entry in dcc-mcp-catalog.yml. Re-run the generator after changing the catalog.

Native node showcase

Willow and grass created through successive generator operations

The willow grows from trunk to primary branches, drooping branches, and leaf cards. The grass adds stalk and frond generators, then edits the width curve to taper the blades. Each step is a new SPM, regenerated by SpeedTree Modeler 10.1.0. The diagram comes from the saved generator IDs and links.

Native output Generators Vertices Faces
Willow 5 7,092 6,166
Grass 3 27,540 59,400

These are procedural geometry studies. The images use unchanged exported OBJ geometry with display materials in headless Blender; they are not Modeler screenshots or native material acceptance. No UI automation builds either asset. Reproduce the workflow · Operation and geometry evidence · Native capability matrix.

Start

This is a source preview. Install the checkout with python -m pip install -e ., then configure the installed Modeler executable:

$env:DCC_MCP_SPEEDTREE_EXECUTABLE = "C:/Applications/SpeedTree/SpeedTree_Modeler.exe"
dcc-mcp-speedtree

The standalone service has its own lifetime and OS-assigned port. Discover it with dcc-mcp-cli --gateway local list, then search and describe its tools. Optional GUI binding uses --pid <PID> --window-handle <HWND> and validates that the window belongs to the named executable. GUI actions must use the project-owned dcc-cua / ui-control route with a fresh exact binding.

Agent workflow

Read the capability audit for the complete tool chain and the limits of modeling, materials, LOD, wind, collision, and cancellation. Use inspect_capabilities for a machine-readable scope report without a GUI.

For a reference-based production candidate, speedtree-production provides production_generate to apply explicit edits to a separate SPM, copy declared and observed texture dependencies, and record a brief and manifest. After an official OBJ export, validate_obj_budget counts triangles using explicit material/category mappings. Both tools report their verification limits: they do not fit reference images, prove material-to-generator ownership, or replace four-view and close-up review by an artist. See the production workflow. The same skill also offers a full-versus-bare OBJ geometry comparison and prepare_bare_project to create a card-removed sibling project for an independent official export. Visualization and image comparison belong to separate downstream tools discovered through the gateway, not this adapter. Record source/export identity, dependency hashes, units and axes, explicit material categories and camera settings in the handoff. Automatic cross-host fitting and visual acceptance remain unverified. Candidate publication requires Windows or Linux with atomic renameat2 support and fails safely elsewhere.

  1. discover_templates and inspect_authoring_catalog enumerate installed generator templates and their stored properties. inspect_graph and inspect_properties read the selected source. edit_graph adds, duplicates, removes, connects, disconnects, renames, hides, or changes properties and existing curve components, writing a new SPM/STT.
  2. discover_content locates installed SPM samples and Games/VFX presets; inspect_preset reports the selected settings.
  3. plan_export checks an SPM, preset, format, and new output directory.
  4. export_batch exports 1–16 SPMs through -export_game or -export, with -export_options. Supported extensions are st9, st, fbx, obj, abc, and usd; availability still depends on product version/license.
  5. export_status reads progress while Core runs the job; Core cancellation stops the owned export process (DELETE /v1/jobs/{id} on the owning service). verify_export checks the recorded file hashes before handoff. Each tree lives in its own directory with its textures and material sidecar.
  6. plan_import prepares a verified handoff. Discover the destination adapter's importer, import into a new destination, and read back geometry, materials, dependencies, transforms, and animation.

The asset bundle contract separates export evidence from target import evidence. Format and target coverage records the tested combinations and feature limits. The service never runs a caller-supplied executable or shell command. Existing output directories are rejected; timeout, failed exit, changed input, missing mesh, or missing explicit material dependency stops the batch and preserves a partial manifest.

Live session with the Modeler (Windows)

The speedtree-live skill observes one exact Modeler window through the project-owned dcc-cua / DCC-MCP ui-control route. live_status reports scoped window metadata and an unverified document name; live_screenshot captures a fresh native PNG. live_revert can deliver its configured physical foreground command, with its application effect reported as unverified.

Capture refusals keep the existing error and message and, when supplied by Core, include its bounded native_evidence diagnostics. These can identify the capture stage, refusal reason and blocker PID/window/bounds without exposing other application text, image data or authorization fields. Missing diagnostics remain absent; a refusal never becomes a successful screenshot.

live_save, live_apply and live_undo are unsupported on the implemented pixels-only adapter. They return error=active_document_binding_unverified before Save input, file/history mutation, reload or output creation. A title basename cannot establish the full document path; the implemented route supplies no reliable document identity or source-bound save confirmation. Caller flags, custom dictionaries, file timestamps, hashes and historical snapshots cannot supply this proof. No automatic retry or synthetic confirmation is used.

live_open is also unsupported on this route. Existing public Core semantic APIs are not a promise of a working Modeler integration: a real provider must be integrated and verified for full document identity, save effects and document opening. The current refusals are a limited safety correction, not completed real-time co-design or a claim that the public protocol lacks these abilities.

All actual UI requires a real Host-authorized, integrity-pinned task bound to an exact PID/HWND. Before its first observation or input, report provider=dcc-cua, the runtime version, PID and HWND. Keep each task on its own issuing connection. There is no raw PowerShell / UI Automation / posted-key / PrintWindow fallback and no authorization from tool JSON or caller environment.

A delivered physical command is not proof of application effect. Partial or unknown completion is not proof of no input, and an additional screenshot is observation rather than save/reload confirmation. The retained transaction engine is not exercised by these unsupported entries. Native Modeler load, save/reload/undo, official export and artistic acceptance remain unverified. See the bootstrap contract.

Unreal

Real SpeedTree source and Unreal Engine result

The same Palm asset in Modeler 10.1 and UE 5.5, using native SpeedTree materials after explicit unit, UV, and texture repairs on a duplicate. The final mesh height is 12.34384 m; collision scale and dynamic wind remain unverified. Capture provenance and repair notes.

Use the installed Unreal ST9 Games preset for a modern Unreal workflow. .spm is the authoring source, not a Content Browser import format. inspect_unreal_project reports descriptor entries only: a missing entry is null because engine plugin defaults may enable it. Check installation and loaded modules through dcc-mcp-unreal; configure missing plugins only after the user authorizes the concrete plan. Configuration and restarts belong to the target adapter. This service does not silently install or enable plugins.

After importing each ST9 via unreal-assets.import_asset, read back the mesh and dependencies. Successful export or enabled plugins alone do not prove a usable imported asset, correct wind, collision, or rendered appearance.

Official references

The repository contains no vendor binaries, licensed sample assets, SDK files, or licensing material. Generated vendor sidecars can contain source-machine paths; keep raw exports and diagnostics local unless reviewed for sharing.

Metadata

Release files for dcc-mcp-speedtree 0.1.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for dcc-mcp-speedtree 0.1.3
File Size Uploaded Provenance Actions
dcc_mcp_speedtree-0.1.3.tar.gz 5.7 MB With provenance Details for dcc_mcp_speedtree-0.1.3.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for dcc-mcp-speedtree 0.1.3
File Interpreter ABI Platform Actions
dcc_mcp_speedtree-0.1.3-py3-none-any.whl Python 3 none any Details for dcc_mcp_speedtree-0.1.3-py3-none-any.whl

Total release size: 5.8 MB

Release files / dcc_mcp_speedtree-0.1.3.tar.gz

Download URL dcc_mcp_speedtree-0.1.3.tar.gz
Size 5.7 MB
Tags Source
SHA-256 checksum
How to use checksums
e5b51ac23fe624aedc69eb1407b4f265a5bf9e16f5d2a184112b7e96aeb61b9f
BLAKE2b-256 checksum
How to use checksums
21ef66eaa1594f78222b81139cb6e036304e02f78ebb5fe58bd4cd317bdb211e
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.

Transparency log

Release files / dcc_mcp_speedtree-0.1.3-py3-none-any.whl

Download URL dcc_mcp_speedtree-0.1.3-py3-none-any.whl
Size 101.2 kB
Tags Python 3
SHA-256 checksum
How to use checksums
bcd6fb6f75aaed2c9b72b6628915912ab81ddf64554899807d6ed0abc3c034a9
BLAKE2b-256 checksum
How to use checksums
aba8c3c73d7014abaa27b5835256cddc908547061a42c234d29c138a6b957300
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.

Transparency log

Security for dcc-mcp-speedtree 0.1.3

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for dcc_mcp_speedtree-0.1.3.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 19, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page