skill

Datasheet Parser Agent

Public Unreviewed

by Kyle Bergstedt

Parses manufacturer PDF datasheets into structured wiki markdown via the ds-extract service, using Claude vision only on ambiguous bbox crops.

Overview Files Skills Discover Packages Releases Dependencies Issues Breadcrumbs Pull Requests Activity
Propose a change
Name Last commit message Last updated
install.sh Add adompkg installer files 1mo ago
package.json Add adompkg installer files 1mo ago
page.json Remove internal identifiers from page.json 10d ago
README.md Initial commit: skill/ds-parser-agent 1mo ago
SKILL.md Add adompkg installer files 1mo ago
uninstall.sh Add adompkg installer files 1mo ago

A Claude Code subagent that parses manufacturer PDF datasheets into structured wiki markdown. Delegates heavy extraction to the ds-extract service (docling + pdfplumber + PyMuPDF) and only uses Claude vision on the ~40 bbox crops the service couldn't resolve autonomously.

What it does

  1. Acquires the PDF (URL, local path, or queue item)
  2. Sends it to ds-extract for deterministic block extraction
  3. Processes the escalation queue (vision on ambiguous crops)
  4. Assembles structured wiki markdown from merged blocks
  5. Publishes to the Adom Wiki with diagrams and screenshots

When to use

Deploy this agent as a background worker for batch datasheet processing, or invoke it directly when you want to parse a specific IC datasheet into the wiki.

Requirements

  • adom-wiki CLI (installed by gallia)
  • ds-queue CLI (installed by gallia)
  • Access to ds-extract service