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 agentmods add skills/cdeistopened/skill-stack/pdf-visionnpx skills add cdeistopened/skill-stack --skill pdf-visiongit clone --depth 1 https://github.com/cdeistopened/skill-stackWhat 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 | $0.00123 | $0.01460 |
| Opus 5 | $0.00062 | $0.00730 |
| Sonnet 5 | $0.00025 | $0.00292 |
| Haiku 4.5 | $0.00012 | $0.00146 |
Grade A, and why
pdf-vision 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 2d 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 — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.
pdf-vision
Vision-powered PDF processing that sees documents the way humans do — not as coordinates and font metadata, but as structured content with meaning.
When to Use
- Converting PDFs to clean markdown (especially scanned, multi-column, or complex layouts)
- Processing documents that pypdf/pdfplumber/Acrobat garble (tables, flowcharts, footnotes)
- Batch processing document archives
- Extracting structured data from government forms, legal contracts, academic papers
- Any PDF task where text-based extraction fails or returns nothing
Quick Start
# Install dependencies
pip install pymupdf google-genai
# Set API key
export GEMINI_API_KEY=your-key
# Analyze a PDF (preflight — no OCR, just document analysis)
python scripts/preflight.py document.pdf
# Convert PDF to markdown
python scripts/ocr_pipeline.py document.pdf
# Convert with custom output path
python scripts/ocr_pipeline.py document.pdf -o output.md
# Convert with specific chunk size
python scripts/ocr_pipeline.py document.pdf -c 8
# Learn from a correction
python scripts/learn.py original.md corrected.md
How It Works
1. Preflight Analysis (~$0.005)
Samples 8 pages from beginning, middle, and end of the document. Sends to Gemini flash-lite to detect: document type, language, column layout, footnotes, tables, scan quality, running headers/footers, text density. Configures the entire pipeline automatically.
Why scattered sampling: A 123-page Latin manuscript with an English preface fools a first-5-pages sample. Sampling beginning + middle + end correctly detects the real document characteristics.
2. Two-Tier Model Routing
Routes documents to the right model based on difficulty:
| Difficulty | Model | Cost/1M tokens (in/out) |
|---|---|---|
| Clean digital | $0.10 / $0.40 | |
| Everything else | $0.25 / $1.50 |
Why two tiers: We tested three models on the same 10 Latin manuscript pages. Gemini 3.1 Flash Lite extracted 4x more content (214K vs 50K chars) than 3.0 Flash Preview, while costing half as much. The cheap model handles clean digital docs fine. Everything else goes to 3.1.
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.
- 2d ago First seen · 133 lines · 123 tokens per session scan A 94513330432b
pdf-vision is a skill published in the GitHub repository cdeistopened/skill-stack (27 stars, last pushed 1mo ago), licensed MIT. It adds 123 tokens to every session and 1,460 once invoked, about $0.0006 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 2026-08-30.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
babysit-pr
Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…
imagegen
Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…
cpu-profile-analysis
Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…