Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add dcc-mcp/dcc-mcp-3dsmax --skill 3dsmax-undogit clone --depth 1 https://github.com/dcc-mcp/dcc-mcp-3dsmaxWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/dcc-mcp/dcc-mcp-3dsmax/3dsmax-undo)<a href="https://agentmods.dev/skills/dcc-mcp/dcc-mcp-3dsmax/3dsmax-undo"><img src="https://agentmods.dev/badge/skills/dcc-mcp/dcc-mcp-3dsmax/3dsmax-undo/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/dcc-mcp/dcc-mcp-3dsmax/3dsmax-undo"><img src="https://agentmods.dev/badge/skills/dcc-mcp/dcc-mcp-3dsmax/3dsmax-undo.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00034 | $0.00909 |
| Opus 5.5 | $0.00014 | $0.00364 |
| Sonnet 5.5 | $0.00007 | $0.00182 |
| Haiku 4.5 | $0.00003 | $0.00091 |
Grade A, and why
3dsmax-undo scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
3ds Max Undo Skill
Reverse 3ds Max operations after the fact. This skill is the safety net for the
adapter's destructive tools: it exposes the host undo and redo stack as typed
MCP tools instead of leaving agents to reach for execute_maxscript.
No silent success
3ds Max exposes no API for the depth of the history stack, so "the host accepted the command" is not the same as "something was undone". Every step is therefore verified:
- A scene fingerprint (node identities, transforms, modifier counts, material bindings, current time, selection) is taken before the step.
- One undo/redo command is executed through a single, capability-detected
channel (
max undo/max redo, falling back to the SDK hold manager). - The fingerprint is taken again. If it is unchanged, the step counts as a no-op and the loop stops.
A request where zero steps changed the scene fails, because the usual cause
is an empty history stack. Callers that expect some operations to leave no
observable trace can pass allow_no_op: true to receive an explicit warning
instead. A request where only some steps changed the scene succeeds with an
explicit warning and reports applied versus requested.
Undo metadata on write tools
Every tool declared destructive: true in this adapter carries an undo:
block stating whether it is reversible and at what granularity. Write paths that
are not destructive may carry the same block when one call touches several
nodes. See docs/UNDO.md for the vocabulary and the per-tool tables.
single_call- oneundo_lastcall reverses the whole tool call.per_node- the host records one entry per node, so passcountequal to the number of nodes the tool reported changing.batch_call- one call wrote N nodes and the host may or may not have grouped them. Undo once, re-read the nodes, and repeat while they still differ. Do not passcount=Nup front: if the host did group the batch, the extra steps consume undo entries that belong to earlier work.script_defined- arbitrary script; the adapter cannot verify coverage.none- not reversible through the host stack (file I/O, scene reset).
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 89 lines · 34 tokens per session scan A f0baddfdbbfb
3dsmax-undo is a skill published in the GitHub repository dcc-mcp/dcc-mcp-3dsmax (9 stars, last pushed today), licensed MIT. It adds 34 tokens to every session and 909 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-20.
Other skills, from other repositories
maya-render
Pipeline stage — render globals, final-frame rendering, and viewport capture: configure render settings, query them, render frames, capture playblasts. Use for producing final or preview imagery. Not for modeling (maya-mesh-ops), animation editing (maya-animation), generic file import/export (maya-geometry), or render…
maya-scripting
Bootstrap stage — escape hatch for Maya work that has no packaged skill yet. Agents should prefer searchskills / dcccapabilitymanifest → loadskill → typed tools (inputSchema + annotations) from domain skills; use executepython or executemel only when no skill matches, for bulk in-process loops, or for API…
maya-import-to-scene
Pipeline stage — structured asset import. Consume an AssetDescriptor produced by maya-asset-source and import the asset (FBX, OBJ, USD) into the current Maya scene via cmds.file(). Handles axis/unit conversion, MaterialMode, PlacementHint, and optional target collection grouping. Returns an ImportToSceneResult with…
dcc-mcp-maya-setup
Set up dcc-mcp-maya for an agent or operator: install Maya Python dependencies with mayapy, generate MCP host configuration, guide the user through loading the Maya plugin, and run a first live-tool smoke prompt.
maya-asset-source
Pipeline stage — asset discovery and resolution. Search local asset libraries, resolve paths to structured AssetDescriptor records, and surface candidate assets for downstream import. Use before maya-import-to-scene to locate what to import.
maya-mock-async
Synthetic async skill for gateway integration tests. Sleeps for N seconds and reports progress every 500 ms. No actual Maya installation required — uses MayaStandaloneDispatcher.