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 mthines/agent-skills --skill video-analysergit clone --depth 1 https://github.com/mthines/agent-skillsWrote 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/mthines/agent-skills/video-analyser)<a href="https://agentmods.dev/skills/mthines/agent-skills/video-analyser"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/video-analyser/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/mthines/agent-skills/video-analyser"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/video-analyser.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 | $0.00155 | $0.05928 |
| Opus 5 | $0.00077 | $0.02964 |
| Sonnet 5 | $0.00031 | $0.01186 |
| Haiku 4.5 | $0.00015 | $0.00593 |
Grade C, and why
video-analyser scanned grade C with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
trap 'rm -rf "$WORK_DIR"' EXIT Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
| Direct video URL | Input matches `^https?://` and does not match `linear\.app` | Download: `curl -fL -o "$WORK_DIR/input.mp4" "$INPUT"`. Use `$WORK_DIR/input.mp4` as `VIDEO_PATH`. | How it starts
The opening of the file, as written. The whole thing — 496 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Video Analyser Skill
Analyse a video file to extract bugs, errors, UI state, and reproduction steps.
The pipeline uses ffmpeg for frame extraction, optional tesseract OCR for text, and optional whisper for audio narration.
The default path (8 keyframes at 768 px) is the Pareto-optimal setting for screen recordings: best quality-per-token on the legibility curve.
By default the selected frames are packed into a single tiled contact sheet and sent to the analysis prompt as one image. This is cheaper on vision tokens and gives a temporal overview at a glance. An opt-in split mode keeps the legacy behaviour of one full-resolution image block per frame — use it when small UI text must be read (see Frame Delivery Mode).
Prerequisites
Required tools
| Tool | Check | If missing |
|---|---|---|
ffmpeg |
which ffmpeg |
Print: ffmpeg is required. Install with: brew install ffmpeg (macOS) or apt install ffmpeg (Linux). Then exit. |
ffprobe |
which ffprobe |
Print: ffprobe is required. It ships with ffmpeg — reinstall ffmpeg. Then exit. |
Optional tools (silent degradation)
| Tool | If present | If absent |
|---|---|---|
tesseract |
Enable OCR mode for text-heavy frames | Skip silently; use vision-only mode |
whisper |
Enable audio transcription when user mentions narration or voiceover | Skip audio step silently |
Run tool detection before any other step.
which ffmpeg || { echo "ffmpeg is required. Install with: brew install ffmpeg (macOS) or apt install ffmpeg (Linux)."; exit 1; }
which ffprobe || { echo "ffprobe is required. It ships with ffmpeg — reinstall ffmpeg."; exit 1; }
OCR_ENABLED=false; which tesseract >/dev/null 2>&1 && OCR_ENABLED=true
AUDIO_ENABLED=false
Minimum ffmpeg version
Require ffmpeg 4 or later.
The select='eq(pict_type\,I)' filter and -vsync vfr were introduced in ffmpeg 4.
Document this requirement in the bail message if detection fails.
Temp Directory
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.
- 11d ago First seen · 496 lines · 155 tokens per session scan C e0b15cbee904
video-analyser is a skill published in the GitHub repository mthines/agent-skills (13 stars, last pushed yesterday), licensed MIT. It adds 155 tokens to every session and 5,928 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it C with 2 findings (recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
debugging-and-error-recovery
Enforces a stop-the-line triage protocol across failure classes — test failures, build breaks, runtime errors, and production incidents — using bisection, safe-fallback design, and instrumentation lifecycle management. Use when any unexpected failure appears and the question is "what class of failure is this and…
go-expert
Use when writing or reviewing Go code that starts goroutines, holds a sync.Mutex/RWMutex, or closes channels - to enforce goroutine lifecycle ownership, lock-hygiene, and channel-close-side rules, and to wire up the detection tooling (go test -race, go vet, go.uber.org/goleak) that catches violations tests alone…
layered-logging-and-alerting-patterns
Guides services where a single logging call fans out to multiple side effects - error tracking (e.g. Sentry), a centralized log-shipping endpoint, and chat/webhook alerting - rather than just writing a log line. Use when adding a log statement to a service with this shape, reviewing why removing a "log statement"…
security-analysis
Runs a two-pass vulnerability scan on a snippet, file, or diff — a static pattern pass across fixed categories (injection, secrets, auth, crypto, deserialization, path traversal, resource exhaustion), then an LLM-reasoning pass for business-logic and race-condition flaws the patterns can't catch. Use for a dedicated…
debug
Runs a tight four-phase loop — reproduce, isolate, diagnose, fix — to root-cause a single bug with evidence, not guesses. Use when given a specific error message, stack trace, or unexpected behavior for one bug and need a fast, mechanical session (reproduce → isolate → diagnose → fix → prevention). For broader…
incident-response-and-postmortems
This skill should be used when the user asks to "write a postmortem", "run an incident retro", "do a blameless postmortem", "document this outage", or is drafting an incident timeline, root cause analysis, or action items after a production incident. Use for the incident-response process and the after-the-fact…