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 howdeploy/deploychan_mcp --skill comfyui-prompt-craftgit clone --depth 1 https://github.com/howdeploy/deploychan_mcpWrote 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/howdeploy/deploychan_mcp/comfyui-prompt-craft)<a href="https://agentmods.dev/skills/howdeploy/deploychan_mcp/comfyui-prompt-craft"><img src="https://agentmods.dev/badge/skills/howdeploy/deploychan_mcp/comfyui-prompt-craft.svg" alt="Measured on agentmods" 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 | $0.00052 | $0.01583 |
| Opus 5 | $0.00026 | $0.00792 |
| Sonnet 5 | $0.00010 | $0.00317 |
| Haiku 4.5 | $0.00005 | $0.00158 |
Grade A, and why
Prompt craft for ComfyUI generation 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 8d 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.
Prompt craft for ComfyUI generation
A prompting approach, practice-tested on Flux and Anima-like models. The examples below are neutral; the technique applies to any content.
Two prompt formats
| Format | When | Example |
|---|---|---|
| Danbooru tags | simple scenes, portraits, standard poses | 1girl, blonde hair, sitting, park |
| Literary text | complex poses, unusual angles, many planes | A girl sits on the edge of a bench, adjusting her hair, seen from above |
Key insight: the literary format instantly solves composition that tags can't handle — the model understands spatial relationships better in free text. Flux is especially fond of literary descriptions along the lines of "a girl stands among bushes in the middle of the street", NOT a comma-separated list of words. Caveat: some LoRA are trained only on tags — test your own stack.
Quality prefix
masterpiece, best quality, newest, highres
Avoid score_5, score_6, year 2025 — these are artifacts of other people's LoRA.
Principles (critical)
- Composition goes in the first 3–4 tags. Angle and pose BEFORE appearance:
profile, from side, sitting on chair→ then hair, clothes. - One tag per concept. Not five synonyms for an angle — just one (
from below). - Negatives DON'T work. The model doesn't understand
no door. Describe what SHOULD be in frame, not what shouldn't. - Remove the conflict, don't compensate. A tag conflicts → delete it. "Fixing" tags bloat the prompt and spawn new conflicts.
- Overloaded → rewrite from scratch. Don't patch a bloated prompt — reset to 15–20 key tags.
- Literary drifts into photorealism → fall back to tags. If 2+ attempts produce photos or break anatomy — return to Danbooru tags (they're more reliable for simple/medium scenes).
- Numbers in tags break the counter.
double bun→hair bun, two buns. Don't mixsoloand1girl.
Anti-patterns and fixes (neutral)
| Problem | Cause | Fix |
|---|---|---|
| Two characters instead of one | double bun (the word "double") or solo+1girl |
hair bun, two buns; remove solo |
| Character with back to the window | front lighting + sunlight from window conflict |
remove front lighting |
| Full body instead of close-up | sneakers, asphalt, full body pull the frame out |
remove shoes/ground, keep close-up |
| Red instead of orange | red hair |
dyed hair, orange hair |
| Character on the couch instead of the floor | sitting on floor, leaning against couch |
sitting on armrest; remove couch |
| Twintails → huge bows | twintails triggers decorative ribbons |
low twintails, small hair tie, elastic hair tie |
| Crouching instead of standing | standing without qualification |
standing straight, upright |
| Photorealism from literary | photo terms: shot, camera, cinematic, overhead |
describe the scene: seen from above, looking down at |
| Name collision in a batch | identical filename_prefix within a second |
unique prefix per job: batch_{i} |
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.
- 8d ago First seen · 124 lines · 52 tokens per session scan A ae9cccf49851
Prompt craft for ComfyUI generation is a skill published in the GitHub repository howdeploy/deploychan_mcp (11 stars, last pushed 9d ago), licensed MIT. It adds 52 tokens to every session and 1,583 once invoked, about $0.0003 per session on Opus 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
comfyui-prompt-engineer
Craft model-specific prompts optimized for the target checkpoint and identity method. Handles FLUX, SDXL, SD1.5, and Wan video models with proper syntax, quality tags, and negative prompts. Use when generating or refining prompts for ComfyUI workflows.
comfyui-prompt-interview
Guided conversational interview to understand a user's creative vision before generating model-appropriate image prompts. Asks clarifying questions about subject, mood, style, and technical preferences (4-7 exchanges), then synthesizes positive prompt, negative prompt, recommended settings table, and pipeline…
flux-txt2img
Build Flux txt2img workflows with Flux.1 Dev (SRPO), Flux 2 Klein 9B, Turbo LoRAs, FluxGuidance, and DualCLIPLoader patterns.
prompt-engineering
ComfyUI prompt engineering knowledge covering CLIP text encoding syntax, weight modifiers, model-specific prompting strategies, and best practices.
seedance-2.0-prompter
Expert prompt engineering for Seedance 2.0. Use when the user wants to generate a video with multimodal assets (images, videos, audio) and needs the best possible prompt.
minimax-h3
Use when writing or debugging prompts for MiniMax H3 (Hailuo 3) video-with-audio generation, running the open weights locally in ComfyUI, choosing a quant or an acceleration LoRA for the VRAM you have, wiring reference-to-video with images, video or audio, or when a generated clip produces gibberish speech, drifts off…