A set of instructions for using FFmpeg and ffprobe to inspect, edit, convert, and verify video and audio files. FFmpeg is a command-line media tool, and ffprobe reads technical details such as duration, frames, and loudness.

It helps cut and join clips, change speed or size, encode for different formats, adjust audio, and add subtitles, text, overlays, or metadata.

It provides task-specific guidance for media operations and checks that rendered files have the expected properties.

ffmpeg — Skill for Claude Code and Codex by dmmdea/offload-harness

Install

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.

Any agent

npx skills add dmmdea/offload-harness --skill ffmpeg

Clone the repo

git clone --depth 1 https://github.com/dmmdea/offload-harness

Made for: Claude Code, Codex.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/dmmdea/offload-harness/ffmpeg/github.svg)](https://agentmods.dev/skills/dmmdea/offload-harness/ffmpeg)

Your own site

<a href="https://agentmods.dev/skills/dmmdea/offload-harness/ffmpeg"><img src="https://agentmods.dev/badge/skills/dmmdea/offload-harness/ffmpeg/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/dmmdea/offload-harness/ffmpeg"><img src="https://agentmods.dev/badge/skills/dmmdea/offload-harness/ffmpeg.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 184 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 980 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

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	$0.00184	$0.00980
Opus 5.5	$0.00074	$0.00392
Sonnet 5.5	$0.00037	$0.00196
Haiku 4.5	$0.00018	$0.00098

Measured 27d ago against content hash f564d433f5fa, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

ffmpeg 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 27d ago.

The scan reads SKILL.md. This mod also ships 5 executable files (reference/ff-ab.ps1, reference/ff9-traps.ps1, reference/ffdump.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

skill/ffmpeg/SKILL.md · 36 lines

How it starts

The opening of the file, as written. The whole thing — 36 lines — stays where its author put it; the contents beside it link to each section on GitHub.

FFmpeg

Drive ffmpeg/ffprobe autonomously and prove every output. The reference folder holds measured facts (2026-09-01, ffmpeg 8.1.2 on the workstation, 8.0.1 on the edge node); read reference/README.md first — its ten rules are the contract — then only the file the task needs.

Routing

Task	Read
Which box / which encoder / how fast	`reference/01-hosts-builds.md` + the live dump JSON for that host
Read a file, count frames, keyframes, loudness, silence, frozen frames, verify a render	`reference/02-probing-and-verification.md`
Cut, trim, `-ss` placement, concat, crossfade, speed, pitch	`reference/03-cutting-concat-retime.md`
Aspect/scale, 16:9, 9:16, thumbnails, GIF/WebP, image sequences, ProRes/DNxHR, delivery recipes	`reference/04-scaling-formats-delivery.md`
Encoder settings, NVENC presets/CQ, CRF, two-pass, GOP, throughput	`reference/05-encoding-nvenc-cpu.md`
Loudness, dead air, ducking, channels, joining audio	`reference/06-audio.md`
Subtitles, drawtext, overlays, mapping, metadata, chapters, timecode	`reference/07-subtitles-text-overlays.md`
Typing a filtergraph in any Windows shell, exit codes, `-progress`	`reference/08-windows-quoting-and-scripting.md`
It "worked" but the output is wrong	`reference/09-failure-modes.md`

Operating rules (short form)

hostname + ffmpeg -version before promising an encoder — the hosts run different major versions (workstation PATH 8.1.2 + 9.0.1 off-PATH for the harness lane; editing rig editing rig 9.0; edge node 8.0.1). Recipes port cleanly, but three behaviours differ: reference/01-hosts-builds.md § Version deltas.
Build commands as argv lists from Python (no shell) or -filter_script; when a shell is unavoidable follow 08. Fonts: only fontfile='C\:/…' (quoted AND colon-escaped) is portable — a drive-less path segfaults from a cwd on the wrong drive, and % in drawtext text needs expansion=none.
Frame-exact cuts re-encode; -c copy only on keyframes. Input -ss resets time — subtitles/enable/-to need -copyts.
Loudness: two-pass loudnorm, then ebur128 re-measure. Dead air: silencedetect → cut list → select/aselect.
Every output passes the 02 verification gate (ffprobe JSON, frame count, loudness, silence, freeze, pixel diff for text) before it is called done. Exit 0 proves nothing.
Test on lavfi clips (02 has the recipes) before touching footage; clean temp files after.
Media judgment (does the video look/sound right) goes through the harness eyes/ears (offload_video_describe, offload_transcribe, offload_vqa) — never a single sampled frame.

Read the full file on GitHub · 36 lines

Files

What ships with it

31 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.

Changes

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.

27d ago Changed f564d433f5fa
1mo ago Changed 7c983e3d2e00
1mo ago First seen · 36 lines · 184 tokens per session scan A 64b4a63cea0c

Subscribe to this mod's changes

ffmpeg is a skill published in the GitHub repository dmmdea/offload-harness (4 stars, last pushed today), licensed Apache-2.0. It adds 184 tokens to every session and 980 once invoked, about $0.0007 per session on Opus 5.5. 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 2026-09-05.

Claude Code Creative and media Gemma Go cli llamacpp local-llm

gh-skill

Manage agent skills with gh skill. Use this skill to discover, preview, install, update, and publish Agent Skills so an agent can self-manage the skills available in its environment.

cli/cli · 41 tokens

printing-press-amend

Amend a published CLI from one of two input sources: (1) dogfood mode mines the active Claude Code session transcript for friction (missing flags, hand- rolled API payloads, silent-null returns); (2) direct-input mode accepts user-supplied asks (rename a command, add commands or feeds, fix a named bug, optionally…

mvanhorn/cli-printing-press · 222 tokens

printing-press-import

Bring a published CLI from the public library into the internal library so it's identical to a freshly-generated copy — module path reverted, manuscripts placed alongside, ready for /printing-press-polish or /printing-press-emboss. Use when the public library has a CLI you don't have locally, or to recover from a…

mvanhorn/cli-printing-press · 104 tokens

printing-press-output-review

Internal sub-skill: agentic review of a printed CLI's sampled command output for plausibility issues that rule-based checks can't encode (substring-match relevance, format bugs, silent source drops, ranking failures). Invoked via the Skill tool by the main printing-press skill at Phase 4.85 and printing-press-polish…

mvanhorn/cli-printing-press · 102 tokens

printing-press-score

Score a generated CLI against the Steinberger bar, compare two CLIs side-by-side.

mvanhorn/cli-printing-press · 22 tokens

octo-drive

Octo Drive — spaces, folders, file upload/download, online-document mounts, share links, invites, IM-attachment transfer. Works with a bot token or a user API key; the CLI routes by token kind. Load after octo-shared.

Mininglamp-OSS/octo-cli · 55 tokens