ComfyUI on a rented GPU: setup, API, Telegram

A workflow for running ComfyUI, a visual system for generating images and videos, on a rented remote GPU server.

In plain words
What is it for?
It guides GPU rental, storage planning, installation, SSH access, ComfyUI API control, workflow execution, and sending generated results to Telegram.
Why use it?
It helps avoid relying on multiple paid generation services when the user wants to operate the models on rented hardware.

Skill for Claude CodeCodex

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.

agentmods
npx agentmods add skills/howdeploy/deploychan_mcp/comfyui-remote
Any agent
npx skills add howdeploy/deploychan_mcp --skill comfyui-remote
Clone the repo
git clone --depth 1 https://github.com/howdeploy/deploychan_mcp

Made for: Claude Code, Codex.

Per session 68 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,405 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 2 findings. Scan, not verified.
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 $0.00068 $0.02405
Opus 5 $0.00034 $0.01203
Sonnet 5 $0.00014 $0.00481
Haiku 4.5 $0.00007 $0.00241

Measured 3d ago against content hash 76f33c7056ca, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ComfyUI on a rented GPU: setup, API, Telegram scanned grade A with 2 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 3d 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

import json, urllib.request, time, random, uuid

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

r = subprocess.run(["curl","-s","-X","POST",
content/skills/comfyui-remote/SKILL.md · 178 lines

How it starts

The opening of the file, as written. The whole thing — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.

ComfyUI on a rented GPU

Instead of a dozen subscriptions and APIs — you rent a GPU server for your work. Under heavy use, per hour it's many times cheaper than generating photos/video on someone else's services. It all comes down to ready-made ComfyUI workflows + setup + deployment time. This skill automates deployment and lets the agent drive generation remotely.

Renting a server (Vast-like hosting)

  1. Billing — top up your balance (card payment).
  2. Template — pick ComfyUI (a preinstalled environment).
  3. Disk slider — models are heavy: a solid model for one task is 20–30 GB, three models to play with is already ~90 GB (plus dependencies, finetunes, LoRA). You pay for disk even when the instance is off (≈$1/day per 100 GB).
  4. Server — for a full day of work take $0.6–1.3/hour. The agent will advise on price/quality. Take a datacenter — otherwise files download for 30+ minutes.
  5. RENT → the instance is created in 0.5–2 min → the Open button.

Inside: Jupyter (to drop in the install script) + Terminal (to run it) + a ComfyUI button. Put scripts in the workspace folder. Shut down → you pay for disk; it's more sensible to delete and bring up a new one next time, if waiting in the queue is expensive.

Keys in .env — the agent asks the user

To download models you need two free keys:

The agent does NOT hardcode keys. It asks the user for them and puts them in .env ON THE INSTANCE:

cp .env.template .env
# HF_TOKEN=<user provides their own>
# CIVITAI_TOKEN=<optional>

.env is not committed to git. Environment variables take priority over .env.

Setup script + models.manifest

Deployment pattern: setup.sh reads .env (env-driven), installs custom nodes (git clone/fetch — idempotent), downloads models per the manifest.

models.manifest — a line-by-line source format:

hf|repo|remote_path|target_dir|target_filename          # from HuggingFace
url|direct_url|target_dir|target_filename|min_bytes      # direct link
civitai|version_id|target_dir|target_filename|name       # from Civitai

All files are skip-if-exists (safe to restart after an interruption). Useful env: HF_TOKEN, CIVITAI_TOKEN, COMFYUI_DIR (/workspace/ComfyUI), SKIP_MODELS, SKIP_CUSTOM_NODES, SKIP_HF_LOGIN, RESTART_COMFYUI.

Read the full file on GitHub · 178 lines

Files

What ships with it

3 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.

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. 3d ago First seen · 178 lines · 68 tokens per session scan A 76f33c7056ca

Subscribe to this mod's changes

ComfyUI on a rented GPU: setup, API, Telegram is a skill published in the GitHub repository howdeploy/deploychan_mcp (11 stars, last pushed 4d ago), licensed MIT. It adds 68 tokens to every session and 2,405 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 2 findings (makes network calls, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

adapter-aws-lambda

Deploy tRPC on AWS Lambda with awsLambdaRequestHandler() from @trpc/server/adapters/aws-lambda for API Gateway v1 (REST, APIGatewayProxyEvent) and v2 (HTTP, APIGatewayProxyEventV2), and Lambda Function URLs. Enable response streaming with awsLambdaStreamingRequestHandler() wrapped in awslambda.streamifyResponse().…

trpc/trpc · 93 tokens

adapter-fetch

Deploy tRPC on WinterCG-compliant edge runtimes with fetchRequestHandler() from @trpc/server/adapters/fetch. Supports Cloudflare Workers, Deno Deploy, Vercel Edge Runtime, Astro, Remix, SolidStart. FetchCreateContextFnOptions provides req (Request) and resHeaders (Headers) for context creation. The endpoint option…

trpc/trpc · 86 tokens

langbot-deploy

Deploy and configure a LangBot instance — Docker / Docker Compose, Kubernetes, the config.yaml model, the Box sandbox runtime, the plugin runtime, and the global API key. Use when installing, deploying, upgrading, or configuring LangBot in production or self-hosted environments. Triggers on "deploy langbot", "langbot…

langbot-app/LangBot · 104 tokens

lepton-cli

Operate NVIDIA DGX Cloud Lepton through the globally installed lep CLI. Use when the user asks to inspect or manage Lepton workspaces, endpoints/deployments, dev pods, batch jobs, fine-tuning jobs, Ray clusters, Slurm clusters, Dynamo endpoints, storage, secrets, nodes, ingress, templates, logs, authentication, or…

leptonai/leptonai · 122 tokens

dstack-prototyping

Use with the dstack skill for model-serving work when the image, serving command, resources, backend/fleet choice, or service behavior is not proven. Guides task-first prototyping on real hardware, choosing fleets/backends that can reuse idle instances and caches, checking vLLM/SGLang sources, and verifying the final…

dstackai/dstack · 80 tokens

modal-serverless-gpu

Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models as APIs, or running batch jobs with automatic scaling.

Orchestra-Research/AI-Research-SKILLs · 42 tokens