New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Personal Clipper

by alecs5am in alecs5am/ralphy

Clip extraction from one long-form source — reads a word-level transcript of a stream, VOD, talk, or video podcast, picks self-contained highlight windows, and cuts each into a short vertical clip…

Apache-2.0Auto-check passedMedia & Creative

Install Personal Clipper

skills CLI
$ npx skills add alecs5am/ralphy --skill personal-clipper -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install alecs5am/ralphy personal-clipper --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/alecs5am/ralphy.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/personal-clipper .claude/skills/personal-clipper && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
personal-clipper
GitHub stars
138
Token cost
~2k tokens
SKILL.md length
980 words
Files
1
Skills in repo
28
Repo updated
First seen
Licence
Apache-2.0

At a glance

Clip extraction from one long-form source — reads a word-level transcript of a stream, VOD, talk, or video podcast, picks self-contained highlight windows, and cuts each into a short vertical clip…

  • Works in 8 steps: Ingest the source. ralphy ref pull… → Transcribe. ralphy ref transcribe… → Select highlight windows (the agent's… → …
  • The user points at ONE existing long-form video and asks for SHORT cuts of it: cut my stream into shorts
  • SKILL.md covers Sub-docs (read on demand), When this mode fires, Source requirements + limits and The flow (one beat at a time,…, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Personal Clipper is an agent skill from alecs5am/ralphy. Clip extraction from one long-form source — reads a word-level transcript of a stream, VOD, talk, or video podcast, picks self-contained highlight windows, and cuts each into a short vertical clip through ralphy clip, then captions, renders, evaluates, and packages the survivors. USE WHEN the user points at ONE existing long-form video and asks for SHORT cuts of it: "cut my stream into shorts", "clip the best moments from this podcast", "make 5 TikToks from this talk", "turn my VOD into clips", "extract the…

Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Media & Creative, covering Influencer and creator marketing. The repository describes itself as: Open-source desktop app for content creation, with an agent runtime and standalone CLI. The licence is Apache-2.0.

When your agent uses it

  • The user points at ONE existing long-form video and asks for SHORT cuts of it: cut my stream into shorts
  • Clip the best moments from this podcast
  • Make 5 TikToks from this talk
  • Turn my VOD into clips

Example prompts

  • “cut my stream into shorts”
  • “clip the best moments from this podcast”
  • “make 5 TikToks from this talk”
  • “/personal-clipper”

Workflow steps

8 steps, taken from the first numbered list in SKILL.md.

  1. Ingest the source. ralphy ref pull (yt-dlp behind the verb) or point at the local file. Never shell out to yt-dlp / ffmpeg directly…
  2. Transcribe. ralphy ref transcribe --language → a word-level transcript (transcript.json). Confirm the language with the user first when…
  3. Select highlight windows (the agent's craft). Read the transcript. Pick self-contained windows: a complete thought with a hook in the…
  4. Cut each window. ralphy clip --from --to --vertical --project per window. --vertical centre-crops to 9:16 (1080x1920); omit it to keep the…
  5. Captions. ralphy generate captions on each clip (Scribe word-level), then bake / overlay per the editor playbook. Snap caption timing to…
  6. Render / bake. ralphy render is the only render path when a clip needs a HyperFrames composition (caption overlay, hook card). A bare…
  7. Evaluate readiness. ralphy project scorecard (#427) for the deterministic ship/repair verdict per clip; /evaluator for a deeper…
  8. Form + distribute Units. ralphy unit create --slug --format video --from "artifacts/videos/.mp4" per surviving clip, then ralphy unit…

What it can do on your machine

Read from SKILL.md and the folder at commit 8d139f0. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Personal Clipper loads about 2k tokens when it runs. Until then it costs about 194 tokens; SKILL.md has 980 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~194
When it runs · the whole SKILL.md, loaded when a task matches
~2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from alecs5am/ralphy at commit 8d139f0, republished under its Apache-2.0 licence (© alecs5am). 980 words, ~2,009 tokens.

Download SKILL.mdSave it as .claude/skills/personal-clipper/SKILL.md (or your agent's skills folder).
name
personal-clipper
description
Clip extraction from one long-form source — reads a word-level transcript of a stream, VOD, talk, or video podcast, picks self-contained highlight windows, and cuts each into a short vertical clip through `ralphy clip`, then captions, renders, evaluates, and packages the survivors. USE WHEN the user points at ONE existing long-form video and asks for SHORT cuts of it: "cut my stream into shorts", "clip the best moments from this podcast", "make 5 TikToks from this talk", "turn my VOD into clips", "extract the highlights". DO NOT FIRE for a long-form overlay edit from audio (that is audio-explainer / podcast-video), or for a generated short with no source footage (that is ugc-review). A brief with no good windows STOPS rather than forcing weak clips.
namespace
user

Personal clipper playbook

Turn one long-form video / stream / podcast into a handful of short vertical clips. The agent reads the source's word-level transcript, picks the strongest self-contained windows, and cuts each into a 9:16 clip through the ralphy clip verb — then captions, renders, evaluates, and packages the survivors. This is the personal-clipper content mode (#436), a supported first-class route. NOT a magic "viral moment detector": the windows are an agent decision grounded in the transcript, and the verb only executes the cut.

Sub-docs (read on demand)

DocWhen to read it
docs/playbooks/modes/personal-clipper.mdThe tight quality floor for the mode (creative objective, gates, negative scope) — read first to set the bar.
.agents/skills/editor/SKILL.mdComposition / caption / render mechanics once a clip is cut.
.agents/skills/editor/references/vo-sync.mdSnapping cut boundaries + caption timing to word-level startMs.
.agents/skills/researcher/references/playbook.mdPulling the source video (ref pull) and frame/transcript tooling.
.agents/skills/audio-explainer/SKILL.mdThe adjacent long-form-OVERLAY mode; contrast with clip-EXTRACTION here.

When this mode fires

A brief that points at one long-form source and asks for short cuts: "cut my stream into shorts", "clip the best moments out of this podcast", "make 5 TikToks from this talk", "turn my 40-minute VOD into clips", "extract the highlights". The deterministic classifier (classifyContentMode) scores these to personal-clipper. It is a SUPPORTED route — promise it.

Contrast with the neighbours:

  • A long-form video built ON TOP of the audio (overlays, faceless explainer) is podcast-video (the audio-explainer skill), not clip extraction.
  • A generated talking-head short is ugc-review / tutorial-ugc, not a cut from an existing source.

Source requirements + limits

  • Source: a video the user owns or has the right to clip (a VOD, stream export, webinar, talk, long podcast with video). Pull a URL with ralphy ref pull <url>; a local file is fine too.
  • Minimum source length: roughly > 3 min — below that, just trim with ralphy clip directly, there is nothing to "select".
  • Clip target duration: 15-90s each (the short-form sweet spot). Default to ~30-60s.
  • Clip count: only as many as the source actually supports. Do NOT pad to a requested number with weak windows (see the stop rule below).
  • Reference gate (AGENTS #3): the source is the user's own footage — no model-reference gate fires. The gate only fires if a generated overlay later introduces a named real entity.

The flow (one beat at a time, checkpoints between)

  1. Ingest the source. ralphy ref pull <url> (yt-dlp behind the verb) or point at the local file. Never shell out to yt-dlp / ffmpeg directly (AGENTS #2).
  2. Transcribe. ralphy ref transcribe <slug> --language <lang> → a word-level transcript (transcript.json). Confirm the language with the user first when non-English.
  3. Select highlight windows (the agent's craft). Read the transcript. Pick self-contained windows: a complete thought with a hook in the first ~2s and a clean out. Each window is a [from, to) pair (word-level startMs → seconds). This is judgement grounded in the transcript text — NOT a detector the verb runs. Present the candidate windows (timestamp + the quoted line) and wait for the user's go before cutting.
  4. Cut each window. ralphy clip <source> --from <ts> --to <ts> --vertical --project <id> per window. --vertical centre-crops to 9:16 (1080x1920); omit it to keep the source aspect. Output lands in <project>/artifacts/videos/ (append-only, auto-versioned). Run the cuts in parallel — they are independent ffmpeg processes.
  5. Captions. ralphy generate captions on each clip (Scribe word-level), then bake / overlay per the editor playbook. Snap caption timing to startMs (AGENTS #16) — never hand-write it.
  6. Render / bake. ralphy render <id> is the only render path when a clip needs a HyperFrames composition (caption overlay, hook card). A bare crop+caption bake can stay an artifacts/videos/ file promoted into a Unit.
  7. Evaluate readiness. ralphy project scorecard <id> (#427) for the deterministic ship/repair verdict per clip; /evaluator for a deeper scroll-stop / hook pass. A failed gate refuses (AGENTS #4) — do not ship over it.
  8. Form + distribute Units. ralphy unit create <id> --slug <clip-slug> --format video --from "artifacts/videos/<clip>.mp4" per surviving clip, then ralphy unit package <id> <slug> (#423) for the platform-spec'd distribution pack. ralphy unit caption for the post copy.
Show full SKILL.md (311 more words)Show less

The "no good clips found" outcome (mandatory)

If the transcript yields no self-contained, hook-bearing windows — a meandering stream with no quotable moments, an interview with no punchy beats, a source that is one long unbroken explanation — STOP and tell the user. Do NOT force weak clips to hit a count. The honest output is "I read the transcript and there are no clips here worth cutting; here is why" plus the closest alternative (e.g. "this reads better as a podcast-video overlay edit" or "give me a richer source"). A handful of strong clips beats a dozen forgettable ones — padding the count is the failure this mode exists to avoid.

ralphy clip flag surface

ralphy clip <source> --from <ts> --to <ts> [--vertical] [--out <path>] [--project <id>]
flagmeaning
<source>Source video (absolute, or relative to cwd).
--from <ts>Window start — seconds (12.5), MM:SS (1:30), or HH:MM:SS (1:02:03).
--to <ts>Window end — same formats. Must be greater than --from.
--verticalCentre-crop the clip to a 9:16 vertical frame (1080x1920). Off = keep source aspect.
--out <path>Explicit output path. Optional when --project is set.
--project <id>Logs the cut to the gen-log and resolves the default --out into <project>/artifacts/videos/.
--force-overwriteSkip the .v2 collision archive (default keeps prior versions).
--note <note>Free-form note recorded in the gen-log row.

The cut is re-encoded for frame-accurate boundaries, so a window chosen from a transcript lands on the spoken word rather than the nearest keyframe.

Common failure modes

  • Treating ralphy clip as a detector — it is not; the agent picks the windows from the transcript and calls the verb per window.
  • Hand-writing timestamps — derive every --from / --to from the transcript's word-level startMs.
  • Padding to a requested clip count — invoke the stop rule instead.
  • Raw ffmpeg / yt-dlp — every step is a ralphy verb (AGENTS #2).
  • Vertical-cropping a clip whose subject lives at the frame edges — a centre crop loses it; keep source aspect (drop --vertical) or reframe in the editor.

© alecs5am, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .agents/skills/personal-clipper of alecs5am/ralphy.

Open the folder on GitHubat commit 8d139f0

Compare with similar skills

Personal Clipper next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Personal Clipper compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Personal Clipper this skillalecs5am/ralphy138—~2kAutomated safety check: PassApache-2.0
Scenario Video Assemblyscenario-labs/skills931—~2.2kAutomated safety check: PassMIT
Product Commercialnodetool-ai/nodetool560—~1.5kAutomated safety check: PassAGPL-3.0
Script Videonodetool-ai/nodetool560—~1.3kAutomated safety check: PassAGPL-3.0
AI Avatar Videoaiskillstore/marketplace4301 repos~2.4kAutomated safety check: PassNone
Ad ReadyLeoYeAI/openclaw-master-skills2.2k—~5.4kAutomated safety check: PassMIT

Similar skills

  • Scenario Video Assembly

    scenario-labs/skills

    A skill your agent uses when generated clips must become a finished video on Scenario via MCP: cutting a shot list together, laying a timeline, concatenating with transitions, overlaying a logo or…

    931 GitHub stars~2.2k tokensUpdated yesterday
    Media & CreativeAuto-check passed

  • Product Commercial

    nodetool-ai/nodetool

    Direct a polished product film or brand commercial in NodeTool.

    560 GitHub stars~1.5k tokensUpdated today
    Media & CreativeAuto-check passed

  • Script Video

    nodetool-ai/nodetool

    Direct a NodeTool video whose timing follows written voiceover, such as an explainer or narrated b-roll.

    560 GitHub stars~1.3k tokensUpdated today
    Media & CreativeAuto-check passed

  • AI Avatar Video

    aiskillstore/marketplace

    Create AI avatar and talking head videos via inference.sh CLI.

    430 GitHub starsUsed in 1 repo~2.4k tokens
    Media & CreativeAuto-check passed

  • Ad Ready

    LeoYeAI/openclaw-master-skills

    Generate advertising images automatically from a product URL + brand profile.

    2.2k GitHub stars~5.4k tokensUpdated 2 mo ago
    Media & CreativeAuto-check passed

  • Ugc Fixloop

    gooseworks-ai/goose-skills

    the UGC fix-loop toolkit — surgically re-render a bad window/beat of a single-take UGC master (stitchreplacement.py, pure FFmpeg) and GPT cross-model review a Seedance prompt before render…

    1.2k GitHub stars~865 tokensUpdated yesterday
    Media & CreativeAuto-check passed

More from alecs5am/ralphy

All 28 skills in this repo

  • Gsap

    alecs5am/ralphy

    GSAP animation reference for HyperFrames. An agent skill from alecs5am/ralphy.

    138 GitHub stars~2.4k tokensUpdated 17 days ago
    Auto-check passed

  • Researcher

    alecs5am/ralphy

    Deep-research workflow for UGC reference material — turns one or more URLs / handles / trend queries into a single cited research report (report.md + sources.json) that a scenarist or art-director…

    138 GitHub stars~1.9k tokensUpdated 17 days ago
    Auto-check passed

  • Editor

    alecs5am/ralphy

    Composition and render craft — assembles scenario.json plus asset-manifest.json into a HyperFrames HTML composition and renders the mp4.

    138 GitHub stars~3.6k tokensUpdated 17 days ago
    Auto-check passed

  • Evaluator

    alecs5am/ralphy

    Quality evaluation of rendered UGC mp4s — scene segmentation, audio loudness / dead-air, caption density, and per-scene visual analysis.

    138 GitHub stars~3.6k tokensUpdated 17 days ago
    Auto-check passed

  • Producer

    alecs5am/ralphy

    End-to-end orchestration — the wrapper that drives the whole production contract across roles, plus batch production.

    138 GitHub stars~4.7k tokensUpdated 17 days ago
    Auto-check passed

  • Scenarist

    alecs5am/ralphy

    Scenario and script craft — writes and reworks the scene-by-scene scenario.json: hook, beat structure, per-scene VO, on-screen text, pacing, and the language/aspect pre-flight.

    138 GitHub stars~3.2k tokensUpdated 17 days ago
    Auto-check passed

Categories

Questions about Personal Clipper

What does Personal Clipper do?

Clip extraction from one long-form source — reads a word-level transcript of a stream, VOD, talk, or video podcast, picks self-contained highlight windows, and cuts each into a short vertical clip…. Personal Clipper is an agent skill from alecs5am/ralphy. Clip extraction from one long-form source — reads a word-level transcript of a stream, VOD, talk, or video podcast, picks self-contained highlight windows, and cuts each into a short vertical clip through ralphy clip, then captions, renders, evaluates, and packages the survivors.

When should I use Personal Clipper?

Personal Clipper fits situations like: the user points at ONE existing long-form video and asks for SHORT cuts of it: cut my stream into shorts; clip the best moments from this podcast; make 5 TikToks from this talk; turn my VOD into clips.

How do I install Personal Clipper in Claude Code?

Run `npx skills add alecs5am/ralphy --skill personal-clipper -a claude-code`. Or copy the skill folder (.agents/skills/personal-clipper in alecs5am/ralphy) into .claude/skills/personal-clipper in your project. Claude Code loads it when a task matches its description.

How do I install Personal Clipper in Codex?

Run `npx skills add alecs5am/ralphy --skill personal-clipper -a codex`. Or copy the skill folder (.agents/skills/personal-clipper in alecs5am/ralphy) into .agents/skills/personal-clipper in your project. Codex loads it when a task matches its description.

Can I use Personal Clipper in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add alecs5am/ralphy --skill personal-clipper -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/personal-clipper, .gemini/skills/personal-clipper, .github/skills/personal-clipper and .opencode/skills/personal-clipper in your project.

What does Personal Clipper need to run?

SKILL.md names no scripts, command-line tools or credentials: Personal Clipper is instructions for the agent only.

Does Personal Clipper access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Personal Clipper safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Personal Clipper use?

Personal Clipper is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Personal Clipper use?

About 2k tokens (SKILL.md is roughly 8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Personal Clipper?

Skills that share tags, products or a category with Personal Clipper: Scenario Video Assembly (scenario-labs/skills, 931 stars), Product Commercial (nodetool-ai/nodetool, 560 stars), Script Video (nodetool-ai/nodetool, 560 stars) and AI Avatar Video (aiskillstore/marketplace, 430 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Personal Clipper?

alecs5am (a GitHub user) maintains it in alecs5am/ralphy, which has 138 GitHub stars. The repository holds 28 skills in this directory. The repository was last updated on September 22, 2026.

Source: alecs5am/ralphy on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.