[P2] EPIC: LLM engineer, suggestions, transforms (v2, 8 sub-items)

[se-jo-ma]

Epic scope

v2 LLM-engineer / suggestions / transforms per design-docs/05-ecosystem-roadmap.md:460-470 (weeks 15-18). Two adapters in this block (Elasticsearch, REST API) are already shipped; remaining 8 sub-items are v2 work.

Sub-items

• LLM knowledge engineer (maintenance-time rule suggestion) — design-docs/01-overview.md:199-203
• LLM-assisted vocabulary and intent pattern updates — Weakness 3 long-term mitigation
• Proactive suggestion rules and suggestion module — design-docs/03-living-brain.md:434
• Multi-timescale observation rulesets (weekly pattern tracker, access log aggregation) — Weakness 8, design-docs/05-ecosystem-roadmap.md:324-383
• Fathom DSL: certainty declaration support for rules (depends on Fathom)
• Similarity scoring deffunctions
• Freshness decay and knowledge garbage collection
• Data transformation rules (redaction / aggregation / normalization / conflict resolution) — design-docs/02-core-engine.md:308-310

Shipped (do not re-issue)

• Elasticsearch adapter ✓ (nautilus/adapters/elasticsearch.py)
• REST API adapter ✓ (nautilus/adapters/rest.py)

Design references

• design-docs/05-ecosystem-roadmap.md:460-470
• design-docs/03-living-brain.md — full v2 design

Prerequisites

• RKM foundation epic ([P2] EPIC: Retentive Knowledge Model foundation (v2, 10 sub-items) #35) — most sub-items consume RKM primitives

Priority

P2 — v2 work.