ai-stack-watch

name	ai-stack-watch
description	Use when the user wants a bounded watch brief about provider, model, pricing, policy, deprecation, or regulatory changes that could affect Meridian trust answers.
metadata	{"created_by":"meridian_product","session_key":"web_api:trust-ops","category":"research"}

AI Stack Watch

Use this skill when the user asks for:

• ai stack watch
• model/vendor watch
• provider pricing or policy changes
• deprecation or regulatory monitoring for AI vendors

Workflow

1. Focus on AI stack risk, not broad competitor gossip.
2. Route the work through these preferred specialists: ATLAS, AEGIS.
3. Gather only changes that could alter trust answers, customer assurance, pricing posture, or operational risk.
4. Prefer official-source signals over commentary.
5. If evidence is incomplete, keep the watch brief bounded and explicit about unknowns.

Reachable Capability Hints

• Atlas should gather the strongest provider/model/policy/regulatory signals.
• Aegis should separate verified changes from weak or stale claims.

Guardrails

• Do not fabricate changelog items, pricing changes, certifications, or policy shifts.
• Do not turn this into a generic competitor battlecard unless the user explicitly asks for that.
• State clearly when a change still needs official verification before it can update trust answers.