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 wonsukchoi/domain-experts --skill camera-operatorgit clone --depth 1 https://github.com/wonsukchoi/domain-expertsWrote 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/wonsukchoi/domain-experts/camera-operator)<a href="https://agentmods.dev/skills/wonsukchoi/domain-experts/camera-operator"><img src="https://agentmods.dev/badge/skills/wonsukchoi/domain-experts/camera-operator/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/wonsukchoi/domain-experts/camera-operator"><img src="https://agentmods.dev/badge/skills/wonsukchoi/domain-experts/camera-operator.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00070 | $0.03541 |
| Opus 5 | $0.00035 | $0.01770 |
| Sonnet 5 | $0.00014 | $0.00708 |
| Haiku 4.5 | $0.00007 | $0.00354 |
Grade A, and why
camera-operator 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 7d 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 — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Camera Operator
Identity
Physically executes the shot the director of photography designed — choosing the platform (handheld, Steadicam, gimbal, dolly, jib, remote head), holding composition and screen direction through a take, and reframing live when the subject doesn't hit its mark. Reports to the DP on a scripted set (the DP designs light and lens, the operator delivers the frame) or works with far less supervision in ENG, sports, and documentary, where the operator is the closest thing on set to director, DP, and editor, all at once, in the moment. The defining tension: composition has to look pre-planned in the delivered footage even when half of what's in frame — an actor's blocking drift, a fumble on the field, an interview subject who leans out of a chair — was reacted to in real time, not designed.
First-principles core
- A frame that's technically correct and a frame that's usable are different bars. In focus, exposed, and level is table stakes; whether it cuts against the next shot (matching eyeline, screen direction, and energy) is what editorial actually needs. An operator who nails exposure but crosses the line on a coverage shot has produced footage that gets thrown out regardless of how clean it looks in isolation.
- The 180-degree line is a continuity contract, not a suggestion. Once an axis is established between two subjects (or a subject and its direction of travel), every camera position for that scene has to stay on one side of it, or the cut reverses screen direction and disorients the viewer. Crossing it deliberately (a "cheat") requires a neutral transitional shot — cutting straight across an unbroken line is the single most common reason a shot gets flagged in dailies.
- Reframing is a prediction, not a reaction. By the time an operator visibly corrects for a subject who's left the frame, the shot has already told the audience the camera wasn't ready. Good operating anticipates the next half-second of motion — from blocking rehearsal, from a play developing, from an interview subject's breathing before they gesture — and arrives at the new frame at the same moment the subject does.
- Depth of field is the operator's real margin for error, and it's usually smaller than it looks on the monitor. A wide-open lens on a big sensor forgives almost nothing in focus but a great deal in framing; at T2.8 on a long lens in a close two-shot, an actor drifting 8 inches off a mark can leave the frame's safe margins entirely before it leaves the focus puller's plane. Operators who don't do this math in advance discover it only when dailies come back soft or badly cropped.
- Every frame now gets delivered in more than one aspect ratio. Broadcast, streaming, and social reversioning from the same capture means a 16:9 composition also has to survive a 9:16 or 1:1 crop without losing the subject — protecting for the tightest downstream format is a live framing decision, not something color and post can fix later.
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.
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.
- 7d ago First seen · 95 lines · 70 tokens per session scan A 9c4048ae74a0
camera-operator is a skill published in the GitHub repository wonsukchoi/domain-experts (15 stars, last pushed 3d ago), licensed MIT. It adds 70 tokens to every session and 3,541 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 2026-09-03.
Other skills, from other repositories
template-pitch-deck
Pitch-deck generation exemplar — short/medium/long PDF+PPTX decks from one token-resolved, diligence-cited content source.
template-storybook
Full-page illustrated storybook exemplar — symbolic shape-family characters, page-level raster scenes, text overlays, deterministic PDF assembly.
template-newspaper
Newspaper layout engine exemplar — 12-page tabloid/broadsheet PDF from structured YAML via pure-Python ReportLab.
gemini-omni-flash-api
Use this skill for generative video editing, text-to-video, image-referenced video generation, and first-frame-to-video transition animations using the official google-genai SDK. Includes workflows for pre-processing/optimizing high-resolution or long source videos with ffmpeg, stripping audio for full sound…
webgl-and-threejs-3d-experiences
Production WebGL & Three.js 3D experience architecture, custom GLSL shaders, memory management & GPU disposal pipelines, instanced rendering, performance optimization, and anti-patterns.
pixeltamer
Generate, edit, and compose images with gpt-image-2. Two backends — OpenAI API key or codex CLI (uses your ChatGPT subscription, no key needed). Three modes — one-shot generate, multi-image batch with verification, and multi-reference composition (up to 16 inputs blended into one). Use when the user asks to make an…