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 agentmods add commands/sandyup/comfyui-mcp/directorgit clone --depth 1 https://github.com/sandyup/comfyui-mcpWhat 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 | $0.00016 | $0.01821 |
| Opus 5 | $0.00008 | $0.00911 |
| Sonnet 5 | $0.00003 | $0.00364 |
| Haiku 4.5 | $0.00002 | $0.00182 |
Grade A, and why
director 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.
This is a copy
97% identical to director — 36 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 135 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/comfy-director — Story-to-Video Production Pipeline
The user wants to create a short film from a story description. This command orchestrates multiple model families (Z-Image, Qwen Edit, WAN FLF) to generate start frames, end frames, and video clips for each scene, then concatenates them into a final video.
Requires the director, z-image-txt2img, qwen-image-edit, and wan-flf-video skills — load them before executing.
Instructions
-
Parse the arguments. The argument is: $ARGUMENTS
- If the argument is
resumeor starts withresume, look for the most recentdirector_state_*.jsonfile in~/code/comfyui-mcp/workflows/and resume from the saved phase. Skip to the Resumption section below. - If the argument is a story/description, proceed with a new project.
- If no argument was provided, ask the user for their story.
- If the argument is
-
Create the state file. Generate a project ID from the current timestamp (e.g.,
story_20260216_143022). Initialize the state file at~/code/comfyui-mcp/workflows/director_state_{project_id}.jsonwith the story text and empty scenes array. -
Phase 1 — Story Planning. Break the story into 2-6 scenes. For each scene, craft:
- description: What happens (1-2 sentences)
- start_prompt: Detailed Z-Image prompt for the opening frame (natural language, descriptive, include camera/lighting/style cues)
- edit_prompt: Qwen Edit instruction to transform the start frame into the end frame
- video_prompt: WAN motion description (include motion verbs: "walks", "turns", "reaches", "camera pans")
Anchor frame strategy: For each scene, decide if the start or end frame is the "anchor" (more complex composition). Generate the anchor first, edit to create the other.
Present the scene plan to the user for approval before proceeding. Update the state file with scene definitions.
-
Ask orientation. Ask the user if they prefer portrait (480x720 video, 832x1472 frames) or landscape (832x480 video, 1472x832 frames). Default to portrait for character-focused stories, landscape for environment/action stories.
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.
- 2d ago First seen · 135 lines · 16 tokens per session scan A ba7fd0e06f18
director is a command published in the GitHub repository sandyup/comfyui-mcp (1 stars, last pushed 1mo ago), licensed MIT. It adds 16 tokens to every session and 1,821 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. It is 97% identical to director, differing in 36 lines, and is treated as a copy.
Other commands, from other repositories
brand-generate
Generate an on-brand document from a saved Brand Profile.
stt
Transcribe a local audio file or remote audio URL into text.
audition-voices
Generate voice audition samples for a character using Venice TTS.
review-video
Make a " Reviews" video — a fast, faceless VO montage of REAL, verified competitor reviews that names the recurring complaints and positions YOUR business as the alternative, then hands off to your own customer testimonials.
transcribe
Transcribe the file at $ARGUMENTS into Markdown using the Frenchie MCP server.
extract-creator-voice
Extract creator brand-voice architecture, on-camera persona, and parasocial-trust contract into Compartment 1.