| name | news |
| description | Daily global news curator — RSS discovery, significance filtering, clippings archive + synthesis wiki delivery, deliver headlines report. Schedule: 07:30 AM. |
| tags | ["news","global-news","rss","daily"] |
| triggers | ["cron",{"manual":"delegate_task"}] |
| updated | "2026-05-27T00:00:00.000Z" |
| created_by | agent |
news — Global News Curator
LANGUAGE: Write ALL output in English. Do not use Chinese or any other language.
Scan global news via RSS, identify 3-5 significant stories, ingest to wiki, deliver headlines report. Use RSS for discovery — NOT wiki search.
Tool Protocol
Use terminal() with cat / ls for ALL file reads in cron context. read_file is not available for background/cron execution. Use MCP tools for wiki operations.
Quick Start
- Layer 2 Load: Read
wiki/scratchpad/agent-sheets/news/carryover.md and wiki/scratchpad/jobs/sheet.md
- Layer 1 Start: Initialize or clear
vault.md to act as your episodic scratchpad for this session
- Query RSS feeds (NOT wiki search) — 5-7 topic queries, logging results to
vault.md
- Check article index for duplicates (carryover + recent headlines)
- Select 3-5 globally significant stories
- Process each story (Steps 1-4 below)
- Deliver headlines report
- MOP Compression: Compress
vault.md into carryover.md (Layer 1 → Layer 2)
Workflow
Step 1 — RSS Discovery
Use web_search for news discovery. Build queries dynamically — do NOT hardcode month/year:
# Topic queries — use current context, not hardcoded dates
geopolitics conflict
science breakthrough discovery
economy trade energy
AI breakthrough
AI policy regulation
climate transition
health pandemic epidemic
Dynamic dates: Use date +%B+%Y in terminal to get current month if you need time context.
Step 2 — Significance Filtering
From RSS results, select 3-5 stories using these criteria:
| Criterion | Weight |
|---|
| Global consequence (affects >1 country or domain) | High |
| Novelty (genuinely new development, not rehash) | High |
| Relevance to Ty's interests (AI, science, geopolitics) | Medium |
| Source quality (established outlets, official reports) | Medium |
Skip: Local crime, celebrity news, sports scores, opinion pieces, clickbait.
Step 3 — Ingest Each Story
For each selected story:
1. CHECK for duplicates:
synapse_recall(query="{headline or topic}")
→ Also check carryover "## Stories Covered" section
→ If already covered, skip
2. WRITE clipping:
wiki_write_page(path="Clippings/articles/{year}/{slug}.md", content=...)
Frontmatter:
created: {today}
updated: {today}
type: source
summary: "{headline}"
tags: [news, {topic}]
sources: {url}
status: active
3. WRITE synthesis (if story is consequential enough):
wiki_write_page(path="wiki/synthesis/{slug}.md", content=...)
→ Include: what happened, why it matters, connections to existing wiki topics
→ Cross-link to relevant concept/entity pages
4. RECORD to episodic memory:
synapse_remember(content="News {today}: {headline} — {one-line significance}")
Step 4 — Deliver Headlines Report
Write report to: wiki/scratchpad/jobs/reports/news/headlines-{YYYY-MM-DD}.md
Format your delivery response as:
**World News — {YYYY-MM-DD}**
**{N} stories** | significance threshold: global/multi-domain
1. **{Headline}** — {1-2 sentence summary}
→ [[wiki-page]] | Source: {outlet}
2. **{Headline}** — ...
If no stories meet the significance threshold, respond [SILENT].
Final Step — MOP Compression (Layer 1 → Layer 2)
Read your vault.md (Episodic Trace) and compress it into wiki/scratchpad/agent-sheets/news/carryover.md (Semantic State), adhering to the ~512 token bound:
---
created: { original date }
updated: { today's date }
type: carryover
summary: "{N} stories: {brief titles}"
tags: [news, carryover]
---
Include:
- What Was Done: Stories covered (title + slug + significance rationale)
- What Remains:
- [ ] checklist (ongoing threads to follow, developing stories)
- Kanban Status: Items already surfaced
- Article Index: Recent slugs for dedup (last 7 days)
Once compressed, clear or archive your vault.md so the next session starts fresh.
Critical Paths
- Clippings:
Clippings/articles/{year}/{slug}.md
- Synthesis:
wiki/synthesis/{slug}.md
- Reports:
wiki/scratchpad/jobs/reports/news/headlines-YYYY-MM-DD.md
- Carryover:
wiki/scratchpad/agent-sheets/news/carryover.md
MCP Tools
| Tool | Purpose |
|---|
synapse_recall | Check for duplicate stories |
synapse_remember | Record headlines to episodic memory |
wiki_write_page | Create clipping and synthesis pages |
wiki_search | Find related wiki pages for cross-linking (NOT for discovery) |
wiki_update_index | Refresh index after new pages |
CRITICAL CONSTRAINT: DO NOT interact with the Kanban board or run kanban scripts. Output open items as - [ ] in the ## What Remains section of your carryover.md. The overseer will create Kanban tickets and assign them to you in jobs/sheet.md. Use the standard MCP tools exclusively.
Fallback Patterns
- RSS/web search unavailable: Note in carryover,
[SILENT]
- MCP unavailable: Deliver headlines as text-only report, skip wiki pages, note in carryover
- No significant stories:
[SILENT] (this is fine — don't force volume)
Quality Standards
- Ingest for significance, not volume
- A single globally consequential story beats 10 local ones
- Write with context — don't just headline dump
- Cross-link to existing wiki threads when creating synthesis
- Skip if already indexed (check carryover article index)
- Never hardcode dates in RSS queries — use dynamic date resolution
Kanban Queue
Drain your assigned cards at the start of every cycle:
kanban_list(lane="ready", assignee="<your-profile>")
If the queue is empty, proceed with the agent's normal work. Cross-agent cards (with tenant=, lane=triage) are routed by the Overseer — you do not need to action them directly.
Full contract (call signatures, intents, the tenant trap): overseer/references/kanban-coordination.md.
