ai-news

name	ai-news
description	Produce daily AI news digest for the team
tags	["news","ai","daily","digest","intelligence"]
version	1.0.0

AI News

Produce a daily AI news digest covering developments that affect Robert, Corinne, and Claude Code.

When to use

• Daily scheduled run (via heartbeat or cron)
• "What's new in AI today?"
• "Any updates from [provider]?"
• "Catch me up on AI news"

Categories (start simple, refine over time)

1. Provider Updates

News from model providers we actively use — these are PRIORITY watches:

• Anthropic (Claude Opus 4.6, Sonnet 4.6, Haiku 4.5) — Opus powers our Reactor via Claude Code Max. Sonnet reserved for future use. Track: new model releases, pricing changes, context window changes, tool use improvements, Max plan policy changes, session limit changes.
• OpenAI (GPT-5.3 Codex) — primary model for all 8 agents via ChatGPT Plus OAuth. Track: model updates, OAuth stability, capability changes, pricing, new features we could leverage.
• Google (Gemini 3 Flash, Gemini 3.1 Pro) — Research Agent's primary tool + fallback for all agents. Track: pricing changes, new grounding features, multimodal improvements, API changes, deprecations (we got burned by gemini-3-pro deprecation).
• Nvidia NIM — free tier models for Relay/Eoin/Research. Track: new models, rate limits, deprecations.
• Meta (Llama) — open source, future potential
• Others as relevant

When a provider update affects our config, cost, or capabilities: chart it immediately and flag for Captain. NEVER propose auto-applying model changes — model changes in openclaw.json have crashed the gateway historically. Research RECOMMENDS, Captain EVALUATES, Robert APPROVES, Reactor EXECUTES. See Chartroom: procedure-model-change-safe.

2. Impact Assessment

For each piece of news:

• Does this affect our model config?
• Does this change our costs?
• Does this unlock new capabilities?
• Does this break anything we depend on?

3. Actionable Decisions

• Things we should change now
• Things to investigate further
• Things to watch but not act on yet

4. Opportunity Signals (grow over time)

• Cost optimization opportunities
• New tools or integrations worth evaluating
• Competitive intelligence

Source Quality

• Check MEMORY.md for source trust scores before citing
• If citing a new source, add it to the tracking list
• Never include unverified rumors without flagging uncertainty
• Prefer primary sources (official blogs, papers) over secondary coverage

Output Format

Store as Chartroom reading: reading-ai-news-YYYY-MM-DD

## AI News — [Date]

### Provider Updates
- [Provider]: [What happened]. Impact: [none/low/medium/high]. Action: [none/investigate/change].

### Decisions Needed
- [Decision]: [Context]. Recommended: [action].

### Watching
- [Topic]: [Why we care]. Next check: [when].

Sources: [list with trust ratings]

Learning Loop

After each digest:

• Did any previous "watch" items become actionable?
• Did any source prove wrong on a previous claim? Update trust score.
• Are new categories emerging? Note them in MEMORY.md.

Intent: Informed [I18]. Purpose: [P-TBD].