用Codex+HyperFrames+Remotion结合一支约 75秒的中文UFO科普视频。 基于GIT上的公开仓库，讲解这个仓库如何把解密 UAP/UFO 文件整理成可读报。 这次我把它们分工成这样： HyperFrames：负责真正出片，用 index.html写画面、文字、转场、动画时间线，然后用 HyperFrames检查画面有没有溢出、对比度够不够，最后渲染成 MP4。你现在看到的成片就是 HyperFrames做出来的。 Remotion：负责视频结构参考，用 Remotion规划每一个镜头多久、怎么接转场、如果以后要转成 React/Remotion 项目应该怎么组织。当前项目里我也放了一个 Remotion参考骨架，但这次没有用Remotion来渲染最终视频。 感兴趣的同学可以看看这个UFO集合仓库： https://github.com/ckpxgfnksd-max/uap-release-analyzer…

ckpxgfnksd-max/uap-release-analyzer

Source: https://github.com/ckpxgfnksd-max/uap-release-analyzer

uap-release-analyzer

A Claude Code / Claude.ai skill that turns a folder of declassified UAP/UFO documents — war.gov “PURSUE” releases, FBI Vault tranches, NARA boxes, AARO publications — into a structured analytic report.

What it does

Run it against a release directory (e.g. ~/Documents/UFO/release_01/) and it produces:

• inventory.csv — one row per file: agency (inferred from filename prefix), document type, page count, size
• text/*.txt — extracted text via pdfplumber, with empty files flagged for the (often majority) of files that are scanned with no text layer
• analytics/
  • top_terms.csv, terms_by_agency.csv — token frequencies
  • entities.json — locations, agencies, phenomena vocabulary, year clusters, names appearing in 5+ files
  • per_file_digest.csv — top terms / locations / redactions / 2-sentence summary per file
  • cross_doc.json — redaction patterns, agency totals, scanned-vs-text split
• REPORT.md — 11-section human-readable analytic writeup

The four scripts are idempotent and incremental — re-running on the same folder skips work that’s already done.

Installation

# Inside Claude Code (per-user skills directory)
git clone https://github.com/ckpxgfnksd-max/uap-release-analyzer.git \
  ~/.claude/skills/uap-release-analyzer

Or package via skill-creator:

python -m scripts.package_skill /path/to/uap-release-analyzer
# produces uap-release-analyzer.skill — install via Claude Code UI

Dependencies: pdfplumber, pypdf. Install via pip install pdfplumber pypdf.

Layout

uap-release-analyzer/
├── SKILL.md              # frontmatter + workflow
├── scripts/
│   ├── inventory.py
│   ├── extract_text.py
│   ├── analyze.py
│   ├── build_report.py
│   └── run_all.py        # convenience: run the four in order
├── references/
│   ├── agency_vocab.md   # filename-prefix → agency rules
│   ├── foia_codes.md     # FOIA exemptions and classification stamps
│   └── war_gov_quirks.md # how war.gov/UFO/ is structured + scraping notes
├── evals/evals.json      # 4 test cases used to iterate the skill
├── ARTICLE.md            # development notes (English)
├── ARTICLE_CN.md         # 中文版开发笔记
└── LICENSE.txt

Usage

# One-shot: full pipeline
python scripts/run_all.py ~/Documents/UFO/release_01/

# Or step-by-step (inventory and extract are the slow parts; both are idempotent)
python scripts/inventory.py    ~/Documents/UFO/release_01/
python scripts/extract_text.py ~/Documents/UFO/release_01/        # all files
python scripts/extract_text.py ~/Documents/UFO/release_01/ 0 25   # chunked
python scripts/analyze.py      ~/Documents/UFO/release_01/
python scripts/build_report.py ~/Documents/UFO/release_01/

Example dataset

The May 2026 war.gov “PURSUE” release this skill was tuned against is mirrored at ckpxgfnksd-max/uap-release-01 (Git LFS, ~2.4 GB / 132 files / 4,157 PDF pages). Clone it as your release_01/ to reproduce the eval scoreboard:

git lfs install   # one-time
git clone https://github.com/ckpxgfnksd-max/uap-release-01.git ~/Documents/UFO/release_01
python scripts/run_all.py ~/Documents/UFO/release_01

Or fetch only the buckets you care about (e.g., the text-bearing DOW mission reports):

GIT_LFS_SKIP_SMUDGE=1 git clone https://github.com/ckpxgfnksd-max/uap-release-01.git ~/Documents/UFO/release_01
cd ~/Documents/UFO/release_01
git lfs pull --include "dow-*"

Video is not mirrored — link to https://www.war.gov/UFO/ for any video material.

Eval scoreboard (iteration-1)

See ARTICLE.md for the build story and the bugs the eval surfaced.

Honest caveats

• Entity extraction is keyword-list + regex, not full NER. Year mentions ≠ incident dates.
• Scanned PDFs (no text layer) produce 0-char .txt files by design — the analyzer treats them as “OCR needed” rather than running OCR (multi-hour). Run Tesseract as a follow-up if you need that content searchable.
• The agency vocabulary is tuned to the May 2026 war.gov tranche. New tranches with new prefixes should be added to references/agency_vocab.md and scripts/inventory.py PREFIX_RULES.

License

MIT. See LICENSE.txt.