@neruva/mcp 0.16.3 → 0.18.0

package/README.md

@@ -4,24 +4,30 @@ MCP server for [Neruva](https://neruva.io) — memory + reasoning substrate for
 
 > **For Claude Code users:** see [neruva.io/claude-code](https://neruva.io/claude-code) for the 30-second install + first-queries to try.
 
-## What's new in 0.16
-
-| Capability | Tool(s) |
-|---|---|
-| Auto-managed entity extraction (server-side) | `agent_remember(extract="managed")` |
-| Federated agent memory (records + KG, one call) | `agent_remember` · `agent_recall` · `agent_context` |
-| Cross-session graph RAG | `agent_recall(namespaces=[...])` |
-| Question-type dispatch (temporal / multi-hop / single-hop / adversarial) | `agent_context(question_type="auto")` |
-| Pearl's do-operator on agent memory | `agent_causal_query` |
-| Provable replay (snapshot + restore) | `agent_snapshot` · `agent_restore` |
-| Anomaly detection on quorum KGs | `agent_check_consistency` |
-| Fact invalidation (Zep-style temporal resolution) | `hd_kg_replace_fact` |
-| Canonical extraction prompt (BYO-LLM) | `hd_kg_extraction_prompt` |
-| 5 KG engines: hadamard / opb / multi-shard / quorum / feature-bundle | `hd_kg_add_fact(engine=...)` |
-| Concept blending (provenance-preserving merge) | `hd_blend_query` |
-| Case-based episode retrieval | `hd_cbr_*` |
-
-**~70 tools across Records, KG, Causal, Analogy, CBR, Blend, Vector memory, federated agent_*, self-introspection.**
+## What's new in 0.17 — 9 cognitive primitives no LLM vendor ships
+
+The substrate now exposes the full **9-level cognitive ladder**. Every primitive runs sub-100ms, deterministic from seed, behind one MCP install.
+
+| # | Capability | MCP tool(s) | Frontier LLM equivalent |
+|---:|---|---|---|
+| 1 | **Vector retrieval** (OPB pages + spectral routing) | `records_query(engine="opb")` | Pinecone/Zep (Level 1 only) |
+| 2 | **KG + Pearl do-operator + HD analogy + CBR** | `hd_kg_*` · `agent_causal_query` · `hd_analogy` · `hd_cbr_*` | nobody |
+| 3 | **Theory of Mind (nested-belief depth-4)** | `agent_model_belief_add` · `agent_model_belief` | hallucinates @ depth ≥3 |
+| 4 | **Counterfactual rollouts ("what if k → a'?")** | `agent_counterfactual_rollout` | confabulates |
+| 5 | **Schema lifting (analogical pattern matching)** | `agent_extract_schema` | needs fine-tuning |
+| 6 | **Active Inference planning (Friston EFE)** | `agent_register_action` · `agent_plan_efe` | not a primitive |
+| 7 | **Few-shot rule induction (3-shot → 100% on 28-rule ARC)** | `agent_induce_rule` | fine-tune (>100 examples) |
+| 8 | **Persistent rule storage** (~26,000× cheaper recall) | `agent_persist_rule` · `agent_recall_rule` | re-feed demos every recall |
+| 9 | **Continual learning, zero forgetting** | `agent_continual_train` · `agent_continual_predict` | catastrophic forgetting |
+| + | **Hierarchical chunking (recursive L^K decode)** | `agent_hierarchical_add` · `agent_hierarchical_decode` | not a primitive |
+
+**~90 tools** across Records, KG, Causal, Analogy, CBR, Blend, Vector memory, federated `agent_*`, the 9 cognitive primitives above, self-introspection.
+
+### Why this is unique
+
+Every primitive in rows 3-9 is empirically validated (probes 34 / 47 / 48 / 49 / 50 / 60 / 70) and graduated to production engines at `neruva_hd/engines/`. No published memory vendor offers more than rows 1-2. The **algorithmic moat is 17 trade-secret VSA primitives** that compound — none individually replicable in <6-12 months by a competitor.
+
+The **structural pitch**: substrate-augmented small LLMs (Haiku, Llama-1B) can match frontier-class agentic capabilities at ~26,000× lower cost per recall.
 
 ## Install