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.
npx skills add damionrashford/media-os --skill media-ocr-aigit clone --depth 1 https://github.com/damionrashford/media-osWrote 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.
[](https://agentmods.dev/skills/damionrashford/media-os/media-ocr-ai)<a href="https://agentmods.dev/skills/damionrashford/media-os/media-ocr-ai"><img src="https://agentmods.dev/badge/skills/damionrashford/media-os/media-ocr-ai/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.
<a href="https://agentmods.dev/skills/damionrashford/media-os/media-ocr-ai"><img src="https://agentmods.dev/badge/skills/damionrashford/media-os/media-ocr-ai.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00209 | $0.03521 |
| Opus 5 | $0.00105 | $0.01760 |
| Sonnet 5 | $0.00042 | $0.00704 |
| Haiku 4.5 | $0.00021 | $0.00352 |
Grade A, and why
media-ocr-ai 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 9d ago.
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.
How it starts
The opening of the file, as written. The whole thing — 245 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Media OCR AI
Context: $ARGUMENTS
Modern open-source OCR. Goes far beyond plain Tesseract by giving four models for four different jobs: PaddleOCR for layout + tables, EasyOCR for "just read the text", Tesseract for the widest language set, TrOCR for handwriting.
For the in-ffmpeg ocr= filter (Tesseract inside a filtergraph, good for logo detection / rough live text), use ffmpeg-ocr-logo. This skill is for offline, high-accuracy, structured OCR — document processing, not video frame grabs.
Quick start
- Read text from a photo (any language): → Step 3 (
ocr.py extract --model easy) - Structured layout of a PDF page (headers, paragraphs, tables): → Step 4 (
ocr.py layout --model paddle) - Cursive handwriting: → Step 5 (
ocr.py handwriting --model trocr) - Multilingual doc (e.g. English + Japanese): → Step 6 (
ocr.py multi-lang --langs en,ja) - Extract a table as CSV: → Step 7 (
ocr.py table --model paddle) - Pre-install a backend: → Step 2 (
ocr.py install <model>)
When to use
- Read text from photos, screenshots, scans, or video frames (single-frame OCR).
- Parse receipts, invoices, business cards, signage — structured documents where layout matters.
- Transcribe handwritten notes, filled forms, or cursive.
- Multilingual content (English mixed with CJK, Arabic, Cyrillic, Thai, etc.).
- PDF text extraction when the PDF is scanned images (use PaddleOCR). For text-native PDFs with embedded font glyphs, reach for
pdftotext/pdfplumberfirst — they're faster and exact. - Do NOT use for video real-time OCR — go to
ffmpeg-ocr-logofor that. Do NOT use for barcode / QR — seeffmpeg-ocr-logo(quirc).
Step 1 — Pick a backend
| Model | License | Strengths | Install difficulty | When to pick |
|---|---|---|---|---|
| PaddleOCR | Apache 2.0 | Best layout analysis + table extraction, 80+ languages | medium (paddlepaddle) | Structured documents, receipts, invoices, tables |
| EasyOCR | Apache 2.0 | Easiest install, good general quality, 80+ languages | easy | "Just read this image" |
| Tesseract 5 | Apache 2.0 | Widest language support (100+), battle-tested, pure CLI | easiest (system pkg) | Legacy, CPU-only, obscure scripts |
| TrOCR | MIT | Transformer trained on IAM handwriting; best cursive by far | medium (transformers) | Handwritten notes, cursive, stylized fonts |
What ships with it
3 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.
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.
- 9d ago First seen · 245 lines · 209 tokens per session scan A aac085335782
media-ocr-ai is a skill published in the GitHub repository damionrashford/media-os (19 stars, last pushed 3mo ago), licensed MIT. It adds 209 tokens to every session and 3,521 once invoked, about $0.0010 per session on Opus 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 2026-09-03.
Other skills, from other repositories
pdf-report
Generate a branded, audience-structured marketing report — executive summary, campaign report, channel deep-dive, competitor report, or monthly/quarterly review — assembled via pdf-generator.py with brand colors, logo, and fonts, and previewed for adjustments before finalizing. Pulls data from campaign-tracker.py…
album-report
Turn an album or a selection into a PDF report: metadata per photo, frames per video, and Claude's own description of what is in each one. Use when the user says "PDF of the album", "album report", "export to PDF", "make a document with these photos", "catalog the album", "identify what is in each video and put it in…
carfax-pdf-review
A review guide for dealer-provided PDFs, including CARFAX vehicle-history reports, service records, and out-the-door price proposals. It extracts accident information, service gaps, and extra financing or dealership charges.
dossier-builder
Use to generate a printed market-research dossier (8-page HTML + PDF) the buyer brings to the dealer at test drive, summarizing buyer profile, regional comps, target OTD, and cross-bid anchors. Triggers include "build dossier", "generate market research PDF", "dealer dossier", "decision document", "make dossier for…
ebook-drive
ค้นหาหนังสือหรือเอกสาร PDF จากแหล่งที่ถูกกฎหมายบนอินเทอร์เน็ต แล้วบันทึกเข้าโฟลเดอร์ Google Drive ของผู้ใช้ โดยอัตโนมัติผ่าน Google Apps Script — ผู้ใช้ไม่ต้องดาวน์โหลดหรืออัปโหลดไฟล์เองแม้แต่ขั้นตอนเดียว ใช้เมื่อผู้ใช้พิมพ์ชื่อหนังสือ/เอกสารพร้อมขอให้หาให้, แนบรูปปกหนังสือแล้วขอให้หา, หรือพูดถึงการเก็บ ebook/PDF…
ebook
ค้นหาและดาวน์โหลดไฟล์ PDF จากแหล่งที่ถูกกฎหมายบนอินเทอร์เน็ต ใช้เมื่อผู้ใช้ต้องการดาวน์โหลดไฟล์ PDF เช่น "หาและดาวน์โหลด Clean Code", "โหลด PDF นี้ให้หน่อย" หรือแค่วาง URL ลิงก์ PDF มาเลย เรียกใช้ผ่าน /ebook เท่านั้น — ไม่ auto-trigger จากบทสนทนา รองรับการตั้งค่า path บันทึกไฟล์และเปลี่ยน path ได้.