R&D Scan Digest
Source: departments/executive/rd-analyst-riley/scans/2026-03-26-scan.md
R&D Scan Digest
Date: 2026-03-26 Analyst: R&D Analyst (Riley)
Discoveries
Evaluate
| # | Discovery | Type | Why It Matters |
|---|---|---|---|
| 1 | MCP Agent-to-Agent Communication (2026 Roadmap) | Protocol/Pattern | MCP roadmap explicitly targets agent-to-agent negotiation, delegation, and multi-step coordination — directly maps to factory's multi-agent architecture. Could formalize how factory workers coordinate. Source |
| 2 | Anthropic 2026 Agentic Coding Trends Report | Knowledge | Anthropic's own report identifies 8 trends including multi-agent orchestration patterns, parallel reasoning across context windows, and scaling agentic coding beyond engineering teams. Factory should benchmark against these. Source |
| 3 | Claude Code source: 'settings' Plugin Inline Declaration |
Feature | Declare plugin entries inline in settings.json — could simplify how factory distributes its plugin ecosystem to client deployments without external marketplace repos. Source |
Watch
| # | Discovery | Type | Re-eval Date | Note |
|---|---|---|---|---|
| 1 | Claude Code Voice Mode (/voice) | Feature | 2026-04-25 | Push-to-talk via spacebar, 20 languages. Could enable voice-driven factory operation for CEO workflow. Not critical path but high convenience factor. Source |
| 2 | Google Colab MCP Server | MCP Server | 2026-04-25 | GPU-powered notebook execution via MCP. Factory doesn't need GPU workloads now, but future production lines (ML-based digital talents) would benefit. Two modes: session proxy + direct kernel. Source |
| 3 | Worktree Sparse Checkout (worktree.sparsePaths) |
Feature | 2026-04-25 | New Claude Code setting for sparse checkout in large monorepos. Factory uses worktrees via superpowers skill — could optimize subagent worktree performance if repo grows. Source |
| 4 | MCP Enterprise Readiness (Auth, Audit, SSO) | Protocol | 2026-04-25 | 2026 roadmap prioritizes enterprise features: audit trails, SSO-integrated auth, gateway behavior. Relevant when delivering MCP-based solutions to enterprise clients (STM). Source |
| 5 | Context Engineering as Discipline | Pattern | 2026-04-25 | Emerging field: systematic construction and management of context for AI systems. Factory already does this implicitly (CLAUDE.md, memory, skills). Worth formalizing if industry converges on standard practices. Source |
Skip
| # | Discovery | Type | Reason Skipped |
|---|---|---|---|
| 1 | Claude Code /loop command | Feature | Already implemented as factory skill (/loop). RD-0010 previously parked this area. |
| 2 | Claude Code /effort command | Feature | Already adopted — SessionStart reminder hook (RD-0008B, score 4.6) covers this. |
| 3 | Claude Code 1M context window | Feature | Already on radar as "Claude Code (Opus 4.6)" in Adopt ring. |
| 4 | Claude Code /color command | Feature | UX cosmetic — no factory impact. |
| 5 | Rich statusline rate_limits field | Feature | Minor UX improvement for rate limit visibility. No action needed. |
| 6 | MCP Elicitation hooks | Feature | New hook types for intercepting responses. Too low-level for current factory needs. |
| 7 | Claude Code $2.5B run-rate revenue | Market | Business metric, not actionable for factory R&D. |
Radar Cross-Reference
No existing radar items need ring changes based on this scan. All Adopt-ring items remain current. The two Assess-ring frameworks (ISO 56002, COBIT APO04) are not impacted by these discoveries.
Potential future radar additions (pending evaluation):
- MCP Agent-to-Agent → could enter Assess ring if evaluation confirms relevance
- Agentic Coding Trends Report → knowledge input, may validate existing Adopt items
Recommended Next Steps
/rd-evaluate "Anthropic 2026 Agentic Coding Trends Report"— Highest priority. Benchmark factory against Anthropic's 8 identified trends. Quick win: confirms factory is ahead or reveals blind spots./rd-evaluate "MCP Agent-to-Agent Communication"— Medium priority. Directly relevant to factory architecture but depends on MCP roadmap timeline. Assess feasibility for factory worker coordination./rd-evaluate "Plugin inline declaration via source:settings"— Lower priority but tactical. Could simplify client deployment packaging. Quick evaluation.