dkyazzentwatwa/chatgpt-skills
3 skills6.5k installs180 starsGitHub
Install
npx skills add https://github.com/dkyazzentwatwa/chatgpt-skillsSkills in this repo
1Ocr Document ProcessorOCR Document Processor is an agent skill for solo builders who need reliable text recovery from images and scanned PDFs inside Claude Code, Cursor, or similar agents. It walks a five-step workflow: choose plain OCR versus structured or specialized parsing, preprocess noisy scans, run the core OCR script, and route receipts or business cards to dedicated scanners. Outputs include searchable PDFs and structured formats agents can chain into validation, indexing, or automation. Guardrails push explicit language selection for accuracy, honest confidence when handwriting or rotation hurts quality, and routing born-digital PDFs to document-converter-suite instead of default OCR. Best when scans, photos, or paper-heavy workflows dominate; skip when sources are already selectable text.4.4kinstalls2Crypto Ta AnalyzerCrypto TA Analyzer appears in the Prism ingest as a finance-oriented agent skill from the chatgpt-skills collection, intended for solo builders who want agent help interpreting cryptocurrency markets. In this catalog snapshot the SKILL.md content did not decode to usable prose, so placement and capabilities are inferred from the skill name and repository context rather than from numbered workflows or install instructions. Treat it as a candidate tag for technical-analysis assistance during live trading iteration until a clean readme is available. Before relying on it for capital decisions, pull the latest skill file from source, confirm invocation triggers, and align outputs with your exchange and compliance constraints. Prism lists it so builders can discover the slug; editorial confidence stays low until documentation is restored.1.5kinstalls3Text SummarizerText Summarizer is a Python-oriented agent skill for extractive summarization: it picks the most important sentences from long documents instead of rewriting them, which keeps citations traceable and reduces hallucination risk when you feed context into Claude Code, Cursor, or Codex. Solo and indie builders use it when competitor research, customer interviews, or internal wikis exceed context windows—you set a 20% ratio, five sentences, or a 100-word cap and get a tight brief for planning or changelog drafts. The skill documents initialization (method and language), in-text and file-based entry points, and bullet-style key-point modes. It fits early research and ongoing content work: same library for a Validate-phase scope doc and a Grow-phase newsletter source dump. It is not abstractive paraphrasing or semantic search; pair it with your own fact-checking before customer-facing copy.699installs