conference-literature-collector

name

conference-literature-collector

description

Collect conference proceedings literature end to end: scrape official program and DBLP/Crossref metadata, download ACM/IEEE PDFs through an isolated per-task browser proxy, record missing papers, screen quantum/QEC candidates, archive yearly PDFs, and produce Chinese research reports. Use when the user asks to collect ASP-DAC/DAC/ICCAD/DATE or similar conference papers, download full texts with local institutional access, organize missing-paper spreadsheets, or find QEC/quantum simulator related work.

Conference Literature Collector

Use this skill for conference-scale literature collection where the output is a local paper library plus metadata, download logs, candidate screening, and related-work notes.

Safety First

Before any publisher or logged-in browser download, use the isolated-browser workflow in references/browser-proxy-isolation.md.

Never change global proxy state for this skill:

• Do not edit Windows system proxy, WinHTTP proxy, browser global settings, or registry proxy keys.
• Do not edit C:\Users\CYL04\.codex\.env for proxy setup.
• Do not edit web-access/config.env for paper downloading.
• Do not attach proxy settings to the user's daily browser process.

The default is a hidden Edge or Chrome instance with a task-specific user data directory, task-specific CDP port, and --proxy-server passed only to that process.

Workflow

1. Create a project root with papers/, metadata/, reports/, logs/, and tools/.
2. Collect metadata from the official conference program first, then enrich with DBLP, DOI, ACM/IEEE, Crossref, or publisher pages.
3. Write a full paper index to metadata/papers.json and metadata/papers.csv.
4. Download PDFs only through authorized routes: publisher access with the user's institution or logged-in isolated browser, open versions such as arXiv/author pages/institutional repositories/Unpaywall, and no paywall bypassing or credential extraction.
5. Record every paper in logs/download_report.csv with status, source URL, local path, and failure reason.
6. Screen quantum/QEC candidates with references/qec-screening.md.
7. Generate reports for candidates, related work, archive layout, and missing-paper summaries.

For ASP-DAC 2023-2026, use references/aspdac-workflow.md and the bundled scripts directly.

Bundled Scripts

• scripts/aspdac_multiyear_pipeline.py: collect ASP-DAC 2023-2025 metadata and QEC candidate reports.
• scripts/publisher_cdp_downloader_bg.js: download ACM/IEEE PDFs through a CDP browser, using CDP_PORT, METADATA_FILE, ONLY_IDS, MAX, and related environment variables.
• scripts/generate_qec_brief.py: generate structured QEC paper briefs from local metadata and extracted text.
• scripts/combine_qec_years.py: combine yearly QEC candidate outputs.
• scripts/build_asp_dac_archive.mjs: build the desktop ASP-DAC archive, missing workbook, and QEC copy set.
• scripts/start_isolated_browser.ps1: start a hidden per-task browser with isolated proxy settings.
• scripts/stop_isolated_browser.ps1: stop only the browser process recorded by the isolated session file.

Handoff To Other Skills

• Use web-access for general web browsing and page inspection when a task does not need a dedicated proxy-isolated publisher browser.
• Use read-qec-paper after this skill identifies or downloads a QEC paper that needs single-paper deep reading.
• Use spreadsheet or document skills only for final artifact formatting when the requested output is .xlsx, .docx, or .pptx.

Output Contract

For a full collection task, produce:

• metadata/papers.json
• metadata/papers.csv
• logs/download_report.csv
• reports/quantum_candidates.md
• reports/qec_simulator_related_work.md
• papers/*.pdf

For archive tasks, produce a year-organized PDF directory, a missing-paper workbook, and a QEC/README.md summary that marks missing full text explicitly.