mvanhorn/printing-press-library

Add GDELT global news intelligence CLI

开放

#1,184 创建于 2026年6月15日

 (1 条评论) (0 个反应) (1 位负责人)Go (590 个派生)github user discovery
enhancementhelp wanted

仓库指标

星标
 (1,953 个星标)
PR 合并指标
 (平均合并 4天 11小时) (30 天内合并 123 个 PR)

描述

Goal

Add a Printing Press CLI for GDELT, focused on global news/event monitoring and entity/topic timelines.

Free source

  • GDELT 2.1 APIs.

Agent jobs

  • Track global coverage volume for a company, person, country, or topic.
  • Build timelines and sentiment/attention curves.
  • Find countries, outlets, and languages driving a story.
  • Provide news-intelligence context without relying on generic search snippets.

Proposed command shape

  • gdelt-pp-cli mentions "OpenAI" --days 7 --agent
  • gdelt-pp-cli timeline "AI regulation" --days 30 --agent
  • gdelt-pp-cli countries "semiconductors" --agent
  • gdelt-pp-cli sources "nvidia" --days 14 --agent
  • gdelt-pp-cli entities "topic" --agent

Non-goals

No paid news APIs, article scraping beyond linked public metadata, or claims that GDELT sentiment is ground truth.\n\n## Implementation path\n\nThis is a new CLI candidate, so do not hand-build it directly in this published library repo. Run it through the canonical Printing Press generation flow in cli-printing-press, keep manuscripts/proofs, then publish into this repo with the publish flow.\n\n## Acceptance criteria\n\n- Research notes explain the source/API/feed contract, rate limits, auth needs, and freshness behavior.\n- CLI exposes a small agent-native command set, not a raw endpoint dump.\n- --json / --agent output is stable and compact.\n- If the source benefits from history, add local SQLite sync plus since/diff commands.\n- Skill includes high-signal recipes and clear non-goals.\n- Dogfood proves at least three realistic agent workflows.

贡献者指南