swallowtail

swallowtail is a skill for Claude Code, Codex from regiellis/godot-mcp-go. It costs 0 tokens per session (13,002 once invoked), scanned A, original, MIT.

A workflow for building and testing games in the Godot game editor through the Swallowtail add-on. Godot is a tool for making games, and the workflow can control the open editor and running game.

In plain words
What is it for?
Use it to create or modify Godot scenes and scripts, run the game, simulate input, inspect behavior, and debug problems. Changes can be undone through Godot's undo system.
Why use it?
It lets you inspect live game state, diagnose failures, change the project, and verify the fix in one loop.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to create or modify Godot scenes and scripts, run the game, simulate input, inspect behavior, and debug problems. Changes can be undone through Godot's undo system.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/regiellis/godot-mcp-go/swallowtail
Install

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.

Any agent
npx skills add regiellis/godot-mcp-go --skill swallowtail
Clone the repo
git clone --depth 1 https://github.com/regiellis/godot-mcp-go

Made for: Claude Code, Codex.

Wrote 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.

agentmods badge for swallowtail

README.md
[![agentmods](https://agentmods.dev/badge/skills/regiellis/godot-mcp-go/swallowtail/github.svg)](https://agentmods.dev/skills/regiellis/godot-mcp-go/swallowtail)
Your own site
<a href="https://agentmods.dev/skills/regiellis/godot-mcp-go/swallowtail"><img src="https://agentmods.dev/badge/skills/regiellis/godot-mcp-go/swallowtail/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.

agentmods 80×15 button for swallowtail

Your own site · 80×15
<a href="https://agentmods.dev/skills/regiellis/godot-mcp-go/swallowtail"><img src="https://agentmods.dev/badge/skills/regiellis/godot-mcp-go/swallowtail.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 0 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 13,002 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00000 $0.13002
Opus 5 $0.00000 $0.06501
Sonnet 5 $0.00000 $0.02600
Haiku 4.5 $0.00000 $0.01300

Measured 2d ago against content hash f52d3cbd7f63, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

swallowtail 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 2d 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.

skills/swallowtail/SKILL.md · 293 lines

How it starts

The opening of the file, as written. The whole thing — 293 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Swallowtail

swallowtail carries one workflow across the running Godot editor, the game process, and the debugger over a WebSocket the Swallowtail editor addon hosts. Discover the live API, create scenes, add nodes, write scripts, play the game, simulate input, inspect state, diagnose failures, apply fixes, and verify the result without leaving the interface. Every editor mutation goes through Godot's UndoRedo, so the user can Ctrl+Z anything.

Prerequisites (check first)

  1. The Godot editor must be open with the swallowtail addon enabled. When a command fails because the editor isn't reachable, run swallowtail status (preflight) rather than guessing. It returns a verdict:
    • running means reachable. Never launch another editor (a second instance stacks and breaks discovery).
    • starting means the process is alive but still booting. Wait a few seconds and retry; do not launch.
    • crashed means a stale discovery file remains but the process is gone. Tell the user it crashed; you may relaunch exactly one editor.
    • closed means no discovery file, so the editor closed cleanly or never started. You may launch exactly one editor if the task needs it.
    • Launch with swallowtail launch [--headless] (run it from inside the project, or pass --project DIR). It applies these same verdict rules itself, starts the editor detached, and sends its output to <project>/.godot/swallowtail-launch.log instead of your terminal, which a raw launch corrupts. It waits until the editor accepts connections and exits non-zero if it never does. Where the CLI is not installed, the fallback is godot --path <project> --editor. Relaunch at most once; if it's still unreachable after one attempt, stop and report rather than looping launches. After a successful launch, status will read running/starting, which is your guard against opening a second.
    • Check project_match too. status reports which project the answering editor serves. When another project's editor holds the port, the verdict reads closed/crashed (nothing is serving this project) with project_match: false and the answering project_path, so follow the verdict and open one for this project. Commands refuse to run in that state rather than editing the other project, which is what the check exists to prevent.
  2. The CLI binary. In this repo: task buildbin/swallowtail.exe. Invoke as bin/swallowtail.exe <group> <command> [--flags] (or task run -- <group> <command> [--flags]). Elsewhere, use the installed swallowtail.
  3. Connection is automatic. The CLI finds the port from <project>/.godot/swallowtail.json (written by the addon) when run from inside the project dir; otherwise pass --port (default 9080).

Read the full file on GitHub · 293 lines

Changes

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.

  1. 2d ago First seen · 293 lines · 0 tokens per session scan A f52d3cbd7f63

Subscribe to this mod's changes

swallowtail is a skill published in the GitHub repository regiellis/godot-mcp-go (61 stars, last pushed today), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 13,002 tokens. 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 2026-09-08.

Related

Other skills, from other repositories

pi-loop-forensics

Diagnose pi-go agent loops and degenerate turns — "agent loop aborted", runaway thinking with no tool calls, repeated phrases. Discriminates genuine model repetition collapse from a race, a tool-parse failure, or a too-low guard, and A/B replays a seed session across providers.

dimetron/pi-go · 65 tokens

aibridge-code-index

Optional read-only AIBridge Code Index lightweight lookup for Unity C# declaration names. Use only when Root Rule declares Code Index enabled and this Skill is installed. Map declaration names to files or positions. Do not use for references, callers, implementations, derived types, diagnostics, comments, strings, or…

liyingsong99/AIBridge · 70 tokens

nightly-session-watch

Nightly sweep of the last 24h of pi-go sessions — anomalous runs, loop aborts, tool error rates, token waste, real prompt-token spend, and whether the observation and palace pipelines are still recording. Triages each finding to the specialist skill that diagnoses it. Use for an unattended daily health check, or on…

dimetron/pi-go · 78 tokens

hz-vr-debug

Debugs Meta Quest and Horizon OS VR/MR applications using the metavr CLI — view logs, capture screenshots, diagnose common issues. Use when troubleshooting crashes, errors, or unexpected behavior on Quest devices.

meta-quest/agentic-tools · 46 tokens

test-fix

Diagnose and minimally fix a failing test, then re-run.

patriceckhart/zot · 17 tokens

hz-perfetto-debug

Analyzes Meta Quest and Horizon OS VR performance using Perfetto traces — frame timing, CPU/GPU bottlenecks, render pass analysis. Use when profiling frame drops, jank, or thermal issues on Quest devices.

meta-quest/agentic-tools · 50 tokens