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 isuzu-shiranui/UnityMCP --skill isuzu-unity-cligit clone --depth 1 https://github.com/isuzu-shiranui/UnityMCPWrote 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/isuzu-shiranui/unitymcp/isuzu-unity-cli)<a href="https://agentmods.dev/skills/isuzu-shiranui/unitymcp/isuzu-unity-cli"><img src="https://agentmods.dev/badge/skills/isuzu-shiranui/unitymcp/isuzu-unity-cli/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/isuzu-shiranui/unitymcp/isuzu-unity-cli"><img src="https://agentmods.dev/badge/skills/isuzu-shiranui/unitymcp/isuzu-unity-cli.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 137 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00073 | $0.02153 |
| Opus 5.5 | $0.00029 | $0.00861 |
| Sonnet 5.5 | $0.00015 | $0.00431 |
| Haiku 4.5 | $0.00007 | $0.00215 |
Grade A, and why
isuzu-unity-cli 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 — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.
isuzu-unity-cli
Each command talks to the Unity Editor that has this project open and prints compact JSON. A call
returns when its work is done: play_mode_play returns once play mode has started, and a call that
takes long waits for its result.
Rules
- Do not list tools. The table below covers common work; for anything else run
isuzu-unity-cli tools --search <words>, thenisuzu-unity-cli tools <name>. - Put every call you already know into one shell command with the template below. Do not read again to confirm a change: each reply already shows the result, and you can answer from it.
- Windows PowerShell deletes double quotes inside arguments, so never pass JSON text. Use
--name value; repeat an option for a list (--paths a --paths b);--values.m_Mass 2sets one field of an object argument; a vector is--position '1,0,2'; anything else goes in a file:--args-file args.json. - Single-quote values with spaces, commas or
@:'/Main Camera','error,warning','@scene:/Boss/Enemy/health'. Unquoted, PowerShell splits1,0,2into three words. - Scene paths start with
/. Siblings that share a name take a 0-based index:/Audio/Speaker[1]. - Property paths are serialized names (
m_Mass,m_LocalPosition.x, a script's field name); the C# name (mass) also works when it is unambiguous, and a miss lists the nearest names.
Template
function u { isuzu-unity-cli @args; if ($LASTEXITCODE) { throw "isuzu-unity-cli $args failed ($LASTEXITCODE)" } }
try {
u call play_mode_play
u call ui_click --text Start
} finally { isuzu-unity-cli call play_mode_stop }
A failed call throws, so the calls after it do not run on a broken state, and finally leaves play
mode. Without play mode, the try/finally is not needed.
Tasks
| Task | Calls |
|---|---|
| Fix compile errors | u verify prints each error with the source lines around it; after the edit, u verify --test --filter <regex> compiles and runs the tests in one call |
| Run tests | u verify --no-compile --test --filter <regex> (EditMode; tests: none matched when nothing matched) |
| Console errors and warnings | u call console_read_logs --type 'error,warning' --limit 30 |
| Find objects | u call scene_browse_hierarchy --name <text> or --component <Type> |
| Objects matching a condition | execute_code with McpSnippet.All<T>(true) and McpSnippet.PathOf (below) |
| Every property of an object | u call inspect_list --object_path <path> --detail full |
| Read or write a property | u call inspect_read --object_path <path> --property_path <prop>; inspect_write also takes --value <v>, or --values.<prop> <v> for several. The component is found from the property; add --component_type <Type> when several have it |
| Change a prefab asset | u call inspect_write --asset_path Assets/X.prefab --property_path <prop> --value <v> saves it; overriddenBy lists scene instances that keep their own value; --object_path Child/Name for a child |
| Create an object | u call gameobject_create --name <n> --parent_path <path> --position '1,0,2' |
| Add a component | u call gameobject_add_component --object_path <path> --component_type <Type> --values.<field> <v> |
| Many objects by a formula | foreach ($i in 0..9) { u call gameobject_create --name "Item$i" --parent_path /Root --position "$($i * 2),0,0"; u call gameobject_add_component --object_path "/Root/Item$i" --component_type Rigidbody --values.m_Mass ($i + 1) }, then u call scene_save |
| Add a script, then use it | one command: write the file, u verify, u call gameobject_add_component --object_path /<name> --component_type <Class> --values.<field> <v>, u call scene_save. A name without a leading / is looked up anywhere in the scene |
| Why a camera does not draw an object | the second snippet below gives every usual cause in one call; fix the one it shows with inspect_write, then u call scene_save |
| Exception in play mode | u call play_mode_play, u call play_mode_step --count 150, u call console_read_logs --type error --stack_trace true --limit 5 |
| A value while playing, or its value after each call that changes it | u call play_mode_play --paused, then u call play_mode_step --seconds 3 --changes --paths '@scene:/<object>/<Component>/<field>': changes lists [frame, time, value] each time the value changed, which is its value right after the code that changed it. No breakpoints or source reading needed |
| Press a uGUI button | u call ui_click --text <label> (or --object_path): textChanges shows the text it changed |
| Save the scene | u call scene_save |
What ships with it
2 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 Changed · +13 lines 7a7062b48ecc
- 1mo ago Changed · -127 lines · -24 tokens per session 72847dfaa975
- 1mo ago Changed · +3 lines 3cf94dbad111
- 1mo ago Changed · -163 lines 8d9ac779bd45
- 1mo ago Changed · +20 lines 072ff3c487c3
- 1mo ago Changed · +50 lines 1188383345c2
- 1mo ago First seen · 328 lines · 97 tokens per session scan A 33e094c8c6ca
isuzu-unity-cli is a skill published in the GitHub repository isuzu-shiranui/UnityMCP (336 stars, last pushed 14d ago), licensed MIT. It adds 73 tokens to every session and 2,153 once invoked, about $0.0003 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 2026-09-05.
Other skills, from other repositories
beehave
Use when using the Beehave addon — pure-GDScript behavior trees with composites, decorators, leaves, a blackboard, and a visual runtime debugger.
godot-debugging
Use when debugging Godot projects — remote debugger, print techniques, signal tracing, common error patterns and fixes.
resolve-diagnostics
Resolves IDE diagnostics (inspections and analyzer findings) at the warning or higher severity level for a given set of files. Use this skill when a workflow's refactoring step calls for resolving diagnostics, or when the user asks to fix warnings, inspections, or static analysis findings in specific files.
cities2-mod-debugging
Use when debugging Cities: Skylines II mod build failures, packaging failures, runtime errors, game logs, UI debugger issues, or mod behavior that does not work in game.
unity-validation
Use when users want to find missing scripts, validate references, or check project health.
unity-console
Use when users want to capture, filter, or clear console logs.