Install
$ agentstack add skill-ravsau-youtuber-skills-youtube-retention ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
YouTube Retention
Input: retention curve points or a description of the video and its average view duration. Data source: if ~/.youtube-skills/metrics/.json exists (see youtube-connect), read retention and curve from it. Else ask the user to paste the curve points from Studio. Output: the curve shape, the problem timestamps, one cutting pattern, and a fix list in priority order.
Steps
- Name the curve shape. Cliff: most of the loss is in the first 30 seconds. Gradual: steady loss. Bump: a spike where viewers rewatch. Flat: holds all the way.
- Compare the average retention with the benchmark table below and label it good, great, or a problem.
- Audit the first 30 seconds: promise delivered by 0:05, no preamble, action started by 0:15, an interrupt at 0:25.
- Mark each timestamp where the curve drops faster than the segments around it. Write what happens on screen there.
- Pick one cutting pattern from the table below for the re-edit.
- If the video runs past 8 minutes, place a payoff just after 8:00.
Benchmarks (average retention)
| Length | Good | Great | Problem | |--------|------|-------|---------| | Under 5 min | 50% | 60% | Under 40% | | 5-15 min | 40% | 50% | Under 30% | | 15-30 min | 30% | 40% | Under 25% | | Over 30 min | 25% | 35% | Under 20% | | Shorts | 70% | 90% | Under 50% |
Cutting patterns
| Pattern | Fits | Rule | |---------|------|------| | Progressive rhythm | Explainer | Fast cuts to 3:00, then slower with b-roll, then bursts after 8:00 | | Contrast | Commentary | Calm 15-25 s cuts, a burst of 5-10 quick cuts every 2-3 min | | Narrative loop | Challenge, story | Restate the premise every 2-3 min | | Hybrid tempo | Tutorial | Cuts every 10-15 s while explaining, holds up to 40 s during examples | | Anchor | Personal, documentary | Cut on emotional beats, hold through full statements |
Output format
## Retention: "[title]"
Shape: [cliff / gradual / bump / flat]. Average: [X]%, [good / great / problem] for [length].
### Drops
1. [m:ss] what happens on screen, why viewers leave
2. ...
### Pattern: [name]. Why: one line.
### Fix list
1. First 30 seconds: [rewrite]
2. [m:ss]: [fix]
3. ...
### Rewatch spike: [m:ss or none]. Clip it as a Short: [yes / no]
Hand the drops and the pattern to youtube-edit-list for the cut rows.
Example
Input: 9-minute video, "MLX vs GGUF on the Same Mac"; 28% average; loses 38% by 0:30, flat to 6:00, bump at 6:40, slow fade after
Shape: cliff, then a bump. 28% is a problem for 5-15 min (under 30%). Drops: 0:00-0:30, the creator explains what MLX is before showing a number. Pattern: hybrid tempo, it is a tutorial. Fix list: 1. open on the two tok/s numbers side by side, then explain. 2. 6:40 has the rewatch spike where the quant setting flips, clip it as a Short. 3. Move the 7:30 summary to 8:10.
CRITICAL: Every fix names a timestamp and one on-screen change. A fix without a timestamp is not a fix.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: ravsau
- Source: ravsau/youtuber-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.