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 dylanroscover/Embody --skill multi-instancegit clone --depth 1 https://github.com/dylanroscover/EmbodyWrote 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/dylanroscover/embody/multi-instance)<a href="https://agentmods.dev/skills/dylanroscover/embody/multi-instance"><img src="https://agentmods.dev/badge/skills/dylanroscover/embody/multi-instance/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/dylanroscover/embody/multi-instance"><img src="https://agentmods.dev/badge/skills/dylanroscover/embody/multi-instance.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00023 | $0.01739 |
| Opus 5.5 | $0.00009 | $0.00696 |
| Sonnet 5.5 | $0.00005 | $0.00348 |
| Haiku 4.5 | $0.00002 | $0.00174 |
Grade A, and why
multi-instance 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 10d 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 — 157 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Multi-Instance Workflow
Envoy supports multiple TouchDesigner instances running simultaneously in the same git repo. Each instance gets its own port, and the bridge can switch between them on demand.
Architecture
Claude Code <--> STDIO Bridge <--> Envoy (TD instance A, port 9870)
|
+---switch---> Envoy (TD instance B, port 9871)
- One bridge process per SESSION (spawned by each AI client's MCP client)
- Per-session pinning - each bridge pins to an instance NAME and re-resolves its port from the registry every tick, so a pinned instance restarting on a new port still self-heals, and a save that renames its
.toeis followed (same TD process). The registryactivefield only seeds NEW bridges without a pin. - A vanished pin fails closed - if the pinned instance leaves the registry (closed, crashed), the session has no target: calls return
envoy.instance.unavailableand nothing is forwarded - never to another instance - until it re-registers (a call that names another instance withinstance=still goes there). Recover withlaunch_td(reopens the pinned.toe) orswitch_instance. - Per-call addressing - every Envoy tool accepts an optional
instanceargument; the bridge routes that one call to the named instance and leaves the pin alone (unknown or unreachable names fail only that call witherror_codeenvoy.instance.unknown/envoy.instance.unreachable). Prefer it over switching when you only need one look at another instance. - Switching is instant and session-local -
switch_instancere-pins THIS session's bridge in-memory; peers are untouched unless you passall_sessions=True(writes the registry default and bumpsactive_epoch, which moves every session) - Registration never re-routes running sessions - a new instance takes the
activedefault slot only when it is vacant or names a dead instance
Instance Registry (.embody/envoy.json)
Each Envoy instance registers itself in .embody/envoy.json at the git root on startup:
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.
- 10d ago Changed · +1 lines 4dff68fff607
- 1mo ago Changed · +1 lines 72a9b170f1ff
- 1mo ago First seen · 155 lines · 23 tokens per session scan A c1e5db77fde3
multi-instance is a skill published in the GitHub repository dylanroscover/Embody (203 stars, last pushed yesterday), licensed MIT. It adds 23 tokens to every session and 1,739 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 2026-08-30.
Other skills, from other repositories
python-api
This skill should be used when writing, editing, or debugging Python code that runs inside TouchDesigner — scripts, expressions, DAT callbacks, Extensions, or one-off snippets passed to executepythonscript — or whenever there is uncertainty about a td class's methods, members, or module-level API. Trigger phrases…
issue_approve
Approve issue to transition to next workflow status.
metagit-cli
CLI-only shortcuts for metagit agents — workspace catalog, discovery, prompts, sync, layout, and config. Use instead of MCP or HTTP API when operating from a shell or agentmode session.
nextcloud-files
Browse and manage Nextcloud files and folders over WebDAV via the nextcloud-agent MCP server — list a directory, read/download a file, upload, create/delete/move/copy, and read properties. Use when the agent must navigate a Nextcloud file tree, fetch a document's bytes, or reorganize storage. Do NOT use for sharing…
x-reader
Read source URLs from X/Twitter, articles, YouTube, Bilibili, podcasts, WeChat, Xiaohongshu, Telegram, RSS, and the open web. Use when the user provides a link and wants the original content, evidence, transcript, summary, or source-grounded analysis.
debug
This skill should be used when the user runs /touchdesigner:debug or asks to debug, investigate, or find TouchDesigner node errors — trigger phrases include "debug my TouchDesigner project", "why is this node broken", "find errors in TD", "investigate node errors", "what's wrong with /project1/geo1", or "fix this…