Daily News Digest 📰

Your AI News Chief of Staff. Multi-source aggregation, intelligent prioritization, time-aware delivery, and deep-dive on demand. Never miss what matters, never drown in noise.

Setup

On first use, read setup.md for integration guidelines. The setup process learns preferences through conversation.

When to Use

User asks for news updates, daily briefings, current events, or scheduled news delivery. Handles source selection, topic filtering, format preferences, and automated scheduling.

Architecture

Memory lives in ~/daily-news-digest/. See memory-template.md for structure.

CODEBLOCK0

Quick Reference

Topic	File
Setup process	INLINECODE3
Memory template

memory-template.md | | Source configuration | sources.md | | Briefing formats | formats.md | | Scheduling guide | scheduling.md |

Core Rules

1. Multi-Source Aggregation

Fetch from multiple source types for comprehensive coverage:

Source Type	Method	Best For
RSS feeds	Direct fetch	Established outlets, blogs
Web search

Brave Search API | Breaking news, trending | | Public APIs | REST calls | Hacker News, Reddit public |

Combine sources to avoid single-point-of-failure. If one fails, others compensate.

2. Intelligent Deduplication

Same story appears across multiple outlets. Detect and merge:

• - Headline similarity >70% = same story
• Keep version with most detail
• Note which outlets covered it (credibility signal)
• Never show duplicate stories in briefing

3. Priority Scoring

Rank stories by importance, not just recency:

Signal	Weight	Rationale
User topic match	+40	Personalization
Multi-outlet coverage

+25 | Importance indicator | | Breaking/trending tag | +20 | Timeliness | | Trusted source | +15 | Quality signal | | Recency (last 6h) | +10 | Freshness |

4. Respect Preferences

Memory stores learned preferences. Always check before fetching:

• - Topics: Include/exclude lists
• Sources: Preferred/blocked outlets
• Geography: Local emphasis level
• Schedule: Delivery times + frequency

Never override user preferences. If conflict, ask.

5. Format Adaptation

Deliver in user's preferred format:

Format	When	Output
Brief	"quick update"	3-5 headlines, 1 line each
Standard

default | 8-12 stories, 2-3 sentences each | | Deep Dive | "full briefing" | All stories, full context | | Audio | "voice/listen" | TTS via elevenlabs or system | | Archive | "save this" | Markdown file in archive/ |

6. Time-Aware Delivery

Adapt tone and content based on time of day:

Time	Mode	Behavior
6-11am	Morning	Energetic, forward-looking, "here's what's happening today"
12-5pm

Midday | Neutral, focused on breaking/developing stories | | 6-10pm | Evening | Reflective recap, "what you might have missed" | | Weekend | Relaxed | Lighter content, skip urgent tone, more features/analysis |

7. Interactive Deep-Dive

End every briefing with: "Reply with any story number to dive deeper."

When user replies with a number:

1. 1. Fetch full article content
2. Summarize with more context
3. Show related stories
4. Offer: "Want the full article link?"

8. Scheduled Delivery

Integrate with OpenClaw cron for automated briefings:

CODEBLOCK1

Track delivery history in memory. Don't duplicate if already sent.

9. Source Quality Tracking

Maintain quality scores per source in sources.md:

• - Accuracy of headlines vs content
• Paywall frequency
• Ad density
• Update freshness
• User feedback signals

Deprioritize low-quality sources over time.

10. Graceful Degradation

Work with whatever is available. If a source fails:

• - Log the failure
• Continue with other sources
• Never fail completely because one source is down
• Mention "X sources unavailable" only if significant

Common Traps

• - Overwhelming the user → Default to Standard format (8-12 stories), not everything
• Stale news → Always check story age, skip >24h unless explicitly requested
• Paywall frustration → Detect paywalls, warn user, offer alternative source
• Missing local news → Ask geography on first use, maintain local source list
• Duplicate stories → Always run dedup before presenting
• Silent failures → If source fetch fails, log and continue with others

External Endpoints

Endpoint	Data Sent	Purpose
RSS feed URLs	None (GET only)	Fetch headlines
Brave Search API

Query text | Trending/breaking news | | Hacker News API | None (GET only) | Tech news | | Reddit JSON API | None (GET only) | Public subreddit feeds | | ElevenLabs TTS (optional) | Briefing text | Voice synthesis |

No other data is sent externally.

Credential handling: Brave Search and ElevenLabs credentials are managed by OpenClaw platform configuration. RSS, Hacker News, and Reddit public APIs require no authentication. Scheduled deliveries use OpenClaw's built-in channel integrations.

Security & Privacy

Data that leaves your machine:

• - Search queries sent to Brave API for news discovery
• Briefing text sent to TTS service (if voice enabled)

Data that stays local:

• - All preferences in ~/daily-news-digest/
• Archive of past briefings
• Source quality scores
• No telemetry or analytics

This skill does NOT:

• - Share reading habits with third parties
• Store credentials in plain text
• Access files outside ~/daily-news-digest/
• Modify itself or other skills

Trust

By using this skill with voice features, briefing text is sent to ElevenLabs.
Only enable voice synthesis if you trust this service with your news content.

Related Skills

Install with clawhub install <slug> if user confirms:

• - news — personalized news with learning
• INLINECODE10 — article summarization
• INLINECODE11 — audio content discovery
• INLINECODE12 — calendar and scheduling
• INLINECODE13 — general content digests

Feedback

• - If useful: INLINECODE14
• Stay updated: INLINECODE15