Xberg is a document-intelligence engine that reads files, URLs, archives, and source trees and extracts text, metadata, images, tables, and structured data, with additional code-language understanding. Developers use it through language bindings, a command-line tool, REST API, or MCP server, and the catalogue entries support those integrations.
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 xberg-io/xberg --skill ocr-pipeline-and-qualitygit clone --depth 1 https://github.com/xberg-io/xbergWrote 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/xberg-io/xberg/ocr-pipeline-and-quality)<a href="https://agentmods.dev/skills/xberg-io/xberg/ocr-pipeline-and-quality"><img src="https://agentmods.dev/badge/skills/xberg-io/xberg/ocr-pipeline-and-quality/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/xberg-io/xberg/ocr-pipeline-and-quality"><img src="https://agentmods.dev/badge/skills/xberg-io/xberg/ocr-pipeline-and-quality.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00053 | $0.00801 |
| Opus 5 | $0.00026 | $0.00400 |
| Sonnet 5 | $0.00011 | $0.00160 |
| Haiku 4.5 | $0.00005 | $0.00080 |
Grade A, and why
ocr-pipeline-and-quality 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 11d 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 — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
OCR pipeline and quality
OCR flows through preprocessing, backend execution, structured conversion, page acceptance, and caching. Backend outputs are not interchangeable measurements.
Backends and execution
- All backends implement
OcrBackend. Tesseract is the default and is bound through the in-repocrates/xberg-tesseractC FFI crate; there is noleptessdependency. OcrBackendTypeisTesseract | PaddleOCR | Candle | Custom. Sceptre is selected by name throughCustom, not a dedicated enum variant.- Run blocking OCR work through
tokio::task::spawn_blocking, minimize runtime/FFI lock duration, and respect backend resource limits. - Check
PageOrientationHandlingbefore assuming a backend handles rotated input. - Validate ISO 639 language codes and required tessdata before execution. Language detection runs after extraction and does not choose traineddata automatically.
Configuration and cache
- Public
types::formats::TesseractConfigand internalocr::types::TesseractConfighave independent defaults. Change both and keep their synchronization test passing. - The OCR cache key combines image hash, backend, config hash, and output format. The config hash includes
TESSERACT_RESULT_SCHEMA_VERSIONand the ordered Tesseract variable set; it contains no build or code identity. - Bump
TESSERACT_RESULT_SCHEMA_VERSIONwhen unchanged image/config inputs can produce different output, or disable the cache for an A/B or revert check. - Default preprocessing is 300 DPI, deskew, and Otsu binarization. Auto-rotation, denoise, contrast enhancement, and color inversion are off unless configured. Native PSM defaults to 3; WASM defaults to 6.
Quality invariants
- Query
confidence_semantics()before interpreting confidence. Never thresholdUncalibratedoutput using a Tesseract-derived scale. - Tesseract font size is typography from hOCR
x_fsize; Sceptre/Paddle font size is a geometric detection-box proxy. Do not compare or threshold them as the same quantity. hocr_font_info=1is required for Tesseract typography; without it font sizes fall back to 12 pt. Sceptre/Paddle do not provide hOCR style fractions.accept_or_reject_ocr_pagecan discard an entire page and its structured paragraphs. Compare accepted pages before word counts. A missing dictionary-invalid ratio is unknown, not zero.- Precision is the scarce resource on the current benchmark corpus. Require an independently grounded F1 A/B for
recall-oriented rewrites; use the
benchmark-workflowskill. - Measure structural output such as headings and lists in Markdown, not Plain.
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.
- 11d ago First seen · 60 lines · 53 tokens per session scan A c16597e2ff2f
ocr-pipeline-and-quality is a skill published in the GitHub repository xberg-io/xberg (9,285 stars, last pushed yesterday), licensed MIT. It adds 53 tokens to every session and 801 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
using-the-mcp-server
Use when converting HTML to Markdown or extracting metadata and tables through the html-to-markdown MCP server's tools, rather than shelling out to the CLI. Covers the tool surface, the auto-installing launcher, and when MCP beats the CLI or SDK.
extraction-pipeline-patterns
Kreuzberg's format detection -> extraction -> fallback orchestration for 75+ file formats.
doc-reader
PDF/DOCX/XLSX/image document intelligence — text extraction, table parsing, OCR, financial statement analysis, contract clause detection, document classification, and format conversion. Use when reading, analyzing, extracting data from, or converting documents.
2slides-ppt-generator
AI-powered presentation generation via the 2slides API — create slides from text, match a reference image style, summarize documents into decks, add AI voice narration, and export pages/audio. Use f.
advogado-criminal
Advogado criminalista especializado em Maria da Penha, violencia domestica, feminicidio, direito penal brasileiro, medidas protetivas, inquerito policial e acao penal.
sf-metadata-parser
Use this skill whenever the user wants to parse, analyze, document, or export SAP SuccessFactors OData metadata ($metadata XML). Triggers include: any mention of 'metadata', '$metadata', 'OData entities', 'entity fields', 'SF metadata', 'SuccessFactors schema', 'data dictionary', or requests to understand, compare, or…