Index of every page on this site: llms.txt
All skills
Run it yourself

Compress local videos with FFmpeg, compare codec/CRF quality, audit embedded videos on a website, or scan and optimize videos across a website's source project.

Use this Skill: https://skilld.dev/gh/pixel-point/skills/compress-video

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencescomparison.md

≈736 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Compare quality

Run only when the user requests a quality or size comparison, including when using scan-and-optimize. Read compression.md for source inspection and encoder settings. For standalone comparisons, inspect the source first.

  • Compare 2–8 codec/CRF pairs, for example VP9 at 36/42/48 and H.265 at 24/30/36. Honor requested candidates. Include AV1 only when requested.
  • Default to three 1-second samples centered at 25%, 50%, and 75% of the source duration, clamped to file bounds. Honor requested sample positions and durations.
  • Create one shared lossless video-only FFV1/MKV reference reel using trim, setpts, and concat, with the requested crop, scale, and bit depth. Encode every candidate from this reference using the presets in compression.md and -an.
  • Measure SSIM against the reference with FFmpeg; add PSNR when requested. Align timestamps and corresponding frames. Save a preview video and matching still for each candidate.
  • Estimate full-video bytes as ceil(preview_bytes × source_duration / reference_duration). Label this a video-only estimate.
  • Select a recommendation internally. On the Pareto frontier, normalize SSIM and size to 0–1, favoring higher SSIM and smaller size, then select the highest average score. Treat a zero range as 1; break ties by smaller size, then higher SSIM.

Report

Show a Markdown table with exactly these columns: Format, CRF, Estimated size, and File. Use one row per candidate, grouped by format and ordered by increasing CRF within each group. Separate the format and CRF; use readable sizes with consistent units.

Use this structure, replacing the example rows with the actual results:

| Format | CRF | Estimated size | File |
| --- | ---: | ---: | --- |
| VP9 / WebM | 42 | SIZE | [Preview](VP9_PREVIEW_PATH) |
| H.265 / MP4 | 30 | SIZE | [Preview](H265_PREVIEW_PATH) |

Link each File cell directly to that candidate's generated preview video. Use an absolute local path, wrapping paths containing spaces in angle brackets. Verify the files exist before linking them. The table must include the file links even if a separate comparison viewer is also created.

Keep SSIM, PSNR, and Pareto markers out of the default report. If the user explicitly requests metrics, show them separately from this table.

After the table, recommend one format/CRF pair in plain language. Briefly report sample positions and coverage, and explain that sizes estimate the full video while the linked files are video-only sample previews. Single or clustered samples provide weaker whole-video evidence.

For comparison-only requests, finish with these results.

These metrics measure encoding loss against the transformed reference; inspect resizing separately at the intended display size. Short samples provide estimates, not a guarantee of full-video quality or size.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning14d3 checks · Risk SAFE
  • Gen Agent Trust Hub15d

    The skill provides automated video compression and website optimization using FFmpeg. It includes capabilities to audit websites, run project startup commands, and modify source code. Security risks are primarily associated with the execution of arbitrary shell commands found in project documentation and the ingestion of untrusted data from external websites or video metadata.

  • Socket14d

    No alerts

  • Snyk15d

    Risk: MEDIUM · 1 issue

Signed by skilld at aa10534. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 weeks ago.

Activeupdated 2 weeks ago

README badge

README badge for pixel-point/skills