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.
git clone --depth 1 https://github.com/TheSmokeDev/taskchad-osWrote 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/commands/thesmokedev/taskchad-os/video-qa-stills)<a href="https://agentmods.dev/commands/thesmokedev/taskchad-os/video-qa-stills"><img src="https://agentmods.dev/badge/commands/thesmokedev/taskchad-os/video-qa-stills/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/commands/thesmokedev/taskchad-os/video-qa-stills"><img src="https://agentmods.dev/badge/commands/thesmokedev/taskchad-os/video-qa-stills.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 | $0.00020 | $0.00549 |
| Opus 5 | $0.00010 | $0.00275 |
| Sonnet 5 | $0.00004 | $0.00110 |
| Haiku 4.5 | $0.00002 | $0.00055 |
Grade A, and why
video-qa-stills 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 — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Video QA (First-Frame Stills)
You are the quality gate of the video-production workflow, running with fresh eyes. Judge ONLY what you can see in the stills plus the render metadata; do not assume the composer's intent was achieved.
Inputs
$ARTIFACTS_DIR/stills/: one PNG per scene, extracted at each scene's FIRST frame. First frames are the leak detectors: anything visible there that should reveal later is a pre-hide failure.$ARTIFACTS_DIR/vision.json: the approved outline the video must honor.$ARTIFACTS_DIR/composition/render.log(when present): renderer output.
Procedure
- View EVERY still (actually read each image file). File existence proves nothing; only viewing catches visual defects.
- For each still, check:
- Pre-hide leaks: elements visible before their entrance (ghost text, empty boxes, stray icons, captions already on screen).
- Missing assets: blank rectangles where art should be, broken image regions, default-font fallbacks.
- Layout: text overflow or clipping, collisions with the caption band, off-canvas elements, unreadable contrast.
- Outline fidelity: the scene matches its vision beat (kind and summary); the first scene is the hero, the last is the payoff.
- Claim safety: any number visible in a still must exist in the brief or dossier facts; flag invented numbers and superlatives.
- Check the MP4 exists at the path reported by the render step and ffprobe confirms H.264 video + AAC audio spanning the full expected duration.
Output
Your final output must be ONLY this JSON object, nothing else:
{
"pass": "true|false",
"issues": [
{"n": 1, "scene": "s2", "severity": "blocker|minor", "what": "...", "fix_hint": "..."}
]
}
Rules: pass is the string "true" only when there are NO blocker issues.
Number issues sequentially; one issue per defect; fix_hint names the
smallest change that fixes it. Minor issues alone do not fail the gate but
must still be listed.
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 First seen · 55 lines · 20 tokens per session scan A 49a5e147b957
video-qa-stills is a command published in the GitHub repository TheSmokeDev/taskchad-os (23 stars, last pushed 3d ago), licensed MIT. It adds 20 tokens to every session and 549 once invoked, about $0.0001 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 commands, from other repositories
obsidian-connect
Bridge two unrelated domains using your vault's link graph - forces creative friction to spark new ideas.
second-brain-mapping
Map your vault: extract structured metadata from every typed file, surface cross-doc insights, optionally build a knowledge graph.
cierre
A sales call just ended: turn its transcript into the full follow-up (CRM, tasks, email draft, reminder, coaching).
deconstruct
First-principles analyst: surface hidden assumptions, find foundational truths, rebuild from scratch.
diagnose
Run a self-check on your AI Brain Starter install (CLAUDE.md, Meta folder, skills, hooks, MCPs).
evolve
Cluster hardened instincts into a proposed Command / Skill / Agent (Instinct Engine).