[Agents/Docs] Consolidate AI files; create authoritative ai/* sources

[ashleyshaw]

AI Ops Summary

Consolidate AI-related documentation into authoritative files under ai/ so contributors and agents have a single, reliable source of truth for agent guidance, model-specific notes, and runner behaviour.

Problem / Opportunity

AI documentation is currently fragmented across multiple files, which increases the risk of duplication, drift, and conflicting guidance. Consolidating these materials into canonical ai/* sources will reduce maintenance overhead, improve onboarding, and make future updates safer and cheaper to maintain.

Approach / Solution

• Author canonical documentation files:
  • ai/agents.md
  • ai/Gemini.md
  • ai/Claude.md
• Retire or replace duplicated AI-related documentation elsewhere in the repo.
• Add ai/RUNNERS.md to document:
  • JS runners
  • Bash runners
  • telemetry hooks
• Update internal references so all AI guidance points to the new authoritative sources.
• Keep the structure minimal and modular to reduce long-term maintenance cost.

Acceptance Criteria

• ai/agents.md authored as the canonical agents reference.
• ai/Gemini.md authored as the canonical Gemini reference.
• ai/Claude.md authored as the canonical Claude reference.
• Duplicate or superseded AI documentation is retired.
• ai/RUNNERS.md documents JS/Bash runners and telemetry hooks.
• Internal references are updated to point to authoritative ai/* sources.

Additional Context

Owner: Agents

Canonical org docs:

• /.github/custom-instructions.md
• /AGENTS.md