# News Sentiment

> Extract structured, source-attributed sentiment context for one symbol.

- Skill: `proompteng/news-sentiment` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add proompteng/news-sentiment`
- Raw SKILL.md: https://api.skillmd.com/api/skills/proompteng/news-sentiment/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: proompteng (https://skillmd.com/u/proompteng)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/proompteng/news-sentiment

---


# news-sentiment

Extract structured, source-attributed sentiment context for one symbol.

## Scripted execution

- Runner: `skills/news-sentiment/scripts/news_run.py`
- Shared lifecycle client: `skills/market-context/scripts/market_context_run_api.py`
- Shared validator: `skills/market-context/scripts/validate_market_context_payload.py`

Use `news_run.py` to:

1. open run lifecycle (`start` + `progress`),
2. validate payload contract,
3. optionally submit parsed evidence rows,
4. finalize to Jangar so DB persistence happens server-side.

## Output contract

Return JSON keys:

- `context_version`
- `as_of_utc`
- `symbol`
- `freshness_seconds`
- `source_count`
- `quality_score`
- `payload.items` (headline, sentiment, source)
- `citations`

## Rules

- Prefer recency over volume.
- Mark missing sources explicitly.

