{
  "id": "page:docs-reference-repos-axtonliu-ai-pair-research",
  "_kind": "Page",
  "_file": "wiki/docs/reference-repos/axtonliu/ai-pair/research.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/reference-repos/axtonliu/ai-pair/research.md",
    "sourceKind": "repo-docs",
    "title": "axtonliu/ai-pair",
    "displayName": "axtonliu/ai-pair",
    "slug": "docs/reference-repos/axtonliu/ai-pair/research",
    "articlePath": "wiki/docs/reference-repos/axtonliu/ai-pair/research.md",
    "article": "\n# axtonliu/ai-pair\n\n- **Archetype**: utility-with-skill\n- **Stars**: 190\n- **Last pushed**: 2026-04-12 (approx)\n- **License**: MIT\n- **Discovered**: 2026-04-12\n- **Source**: gh-search (code: SKILL.md)\n- **Skills found**: 1\n\n## Summary\nAI Pair Collaboration skill. Coordinates multiple AI models to work together: one creates (Author/Developer), two others review (Codex + Gemini). Works for code, articles, video scripts, and creative tasks. Uses Claude Code's native Agent Teams capability with heterogeneous model reviewers.\n\n## Assessment\nMEDIUM-HIGH VALUE. The multi-model review pattern (author + 2 reviewers from different model families) is a sophisticated orchestration pattern. The insight that \"different AI models have fundamentally different review tendencies -- they look at completely different dimensions\" is valuable. This maps directly to babysitter's multi-harness capabilities.\n\n## Extraction Priority\nMEDIUM-HIGH -- The heterogeneous-reviewer pattern is directly extractable:\n- Multi-model review -> methodologies/ (quality assurance methodology)\n- Team coordination -> specializations/shared/ (cross-domain collaboration pattern)\n\n## Processes\n1. **heterogeneous-ai-review** -- Author creates -> Codex reviews from angle A -> Gemini reviews from angle B -> synthesize feedback\n2. **ai-dev-team** -- Developer + codex-reviewer + gemini-reviewer for code\n3. **ai-content-team** -- Author + codex-reviewer + gemini-reviewer for content\n\n## Plugin Ideas\n- **multi-harness-review plugin**: Babysitter plugin that dispatches review tasks to multiple harnesses and synthesizes their feedback\n\n## Library Mapping\n\n| Extractable Process | Library Status | Action | Existing Path | Target Placement |\n|-------------------|----------------|--------|---------------|------------------|\n| Heterogeneous AI Review | NEW | Multi-model review pattern with feedback synthesis | - | methodologies/heterogeneous-ai-review/ |\n| AI Development Team Coordination | NEW | Code review with multiple AI model perspectives | - | specializations/shared/ai-team-coordination.js |\n| AI Content Team Orchestration | NEW | Content review and creation with diverse model feedback | - | specializations/shared/ai-content-team-orchestration.js |\n\n## Plugin Marketplace Mapping\n\n| Plugin Idea | Marketplace Status | Action | Existing Plugin | Target Placement |\n|-------------|-------------------|--------|-----------------|------------------|\n| Multi-Harness Review | NEW | Orchestrate review tasks across multiple AI harnesses with feedback synthesis | - | plugins/a5c/marketplace/plugins/multi-harness-review/ |\n\n## Implicit Procedural Knowledge\n- Different model families find different categories of issues (coverage maximization through diversity)\n- Team architecture: Commander (user) -> Team Lead (Claude) -> Specialists (Codex, Gemini)\n- team-stop command for clean resource teardown\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-reference-repos",
      "to": "page:docs-reference-repos-axtonliu-ai-pair-research",
      "kind": "contains_page"
    }
  ]
}