AERO-WebScraper 0.2.0 — skills-only

AERO-WebScraper is now an instruction-based research and collection plugin. ChatGPT performs the browsing, source inspection, extraction, classification, and reporting with its available native tools. There is no custom scraper, Python application, MCP server, API key, database service, local installation, or externally hosted backend to run.

Use it

Select AERO-WebScraper in ChatGPT and ask, for example:

Collect Nintendo Wii hardware and networking documentation. Prioritize primary references. Inspect up to 20 relevant sources, extract technical notes, organize a source catalog, and create a checkpoint. Clearly separate saved originals from links.

Resume this collection using the attached CHECKPOINT.json and CATALOG.jsonl. Focus on gaps in replacement-service protocol documentation.

The existing default prompt remains a planning request. Ask explicitly to collect or research to execute a collection instead of only drafting a plan.

Eight skills

AERO-WebScraper coordinates. HardwareDocs, Firmware, DevLibraries, CodeResearch, GamesHomebrew, and NetworkServices provide specialist research procedures. AERO-Library maintains the shared catalog, notes, provenance, exports, and checkpoints. These are instructions used by one assistant, not eight running processes.

Outputs

A collection normally produces REPORT.md, CATALOG.jsonl, CHECKPOINT.json, and SOURCE_NOTES.md. With native file support, it may include actual original files and a ZIP. Without file creation it returns the same content inline and reports that no file was created. Unsupported binary downloads remain source links, not fake saved files.

Boundaries

Tool availability, source accessibility, copyright limits, quotas, and host permissions still apply. Skills cannot guarantee complete website mirrors, continuous background crawling, immutable storage, or arbitrary downloads. Catalog IDs and retrieval preferences are instruction-level bookkeeping, not backend enforcement. Persistence is reported only for confirmed saves; temporary working files are not a cross-session database.

The active repository contains only instructions, references, templates, and manifests. The 0.1.0 Python implementation remains in Git history, not in the active plugin. See deployment, limitations, validation, and migration.