Improve graph rendering and context efficiency

README.md

@@ -157,7 +157,7 @@ selection, set `ENGRAPHIS_UPDATE_EXTRAS` to a comma-separated list (for example
 `server,mcp`), or set it to `none` for the base package only.
 
 > **Upgrading to 1.4:** `engraphis-mcp` now exposes the nine-tool Smart gateway. Integrations that
-> require the former 34 direct tool names should run `engraphis-mcp-classic`. The SQLite schema
+> require the former 35 direct tool names should run `engraphis-mcp-classic`. The SQLite schema
 > in the 1.4.0 release was version 9. Existing v7-to-v8 databases already contain `confidence`
 > and `pinned_at`/`unpinned_at`; v9 adds the `memory_tombstones` repository-scope column/table
 > and performs a one-time entity-canonicalization repair, then migrates automatically on first
@@ -387,7 +387,7 @@ the indicated read or action executor; no profile selection is required. The gat
 the discovered capability again before it runs it, and clients remain responsible for their
 normal destructive-action approval boundary.
 
-Existing clients that pin the historical 34 named tools can use
+Existing clients that pin the historical 35 named tools can use
 `engraphis-mcp-classic` (or `engraphis-mcp-http --classic`). The complete classic inventory,
 including `engraphis_check_update`, is in the [MCP tool reference](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/MCP_TOOLS.md).
 
@@ -562,7 +562,7 @@ when you are ready to evaluate the service boundary and billing options.
 | | Free (available now) | Pro: $10/mo or $100/yr | Team: $20/seat/mo or $200/seat/yr |
 |---|---|---|---|
 | Dashboard WebUI (with built-in inspector) | ✓ | ✓ | ✓ |
-| Memory engine + Smart MCP (Classic 34-tool compatibility) | ✓ | ✓ | ✓ |
+| Memory engine + Smart MCP (Classic 35-tool compatibility) | ✓ | ✓ | ✓ |
 | Version-chain diffs, offline knowledge graph | ✓ | ✓ | ✓ |
 | Manual local consolidation (dry-run by default) | ✓ | ✓ | ✓ |
 | Local workspace export (portable v2 JSON: memories, source manifests, graph/code evidence, sessions, audit, and receipts) | ✓ | ✓ | ✓ |
@@ -580,7 +580,7 @@ when you are ready to evaluate the service boundary and billing options.
 
 ## MCP tools
 
-Engraphis exposes a zero-configuration Smart MCP gateway plus a 34-tool Classic compatibility
+Engraphis exposes a zero-configuration Smart MCP gateway plus a 35-tool Classic compatibility
 server across memory, recall, code graphs, governance, sessions, and privacy-safe audit receipts.
 The focused [MCP tool reference](https://github.com/Coding-Dev-Tools/engraphis/blob/main/docs/MCP_TOOLS.md) is the source for
 the full inventory and parameters.
@@ -758,7 +758,7 @@ engraphis/
 │   ├── backends/            # pluggable embedder / vector index / reranker / codegraph / sync transports / encryption
 │   ├── factory.py           # outer v2 composition root; selects and injects concrete backends
 │   ├── service.py           # validated MemoryService facade
-│   ├── mcp_server.py        # Smart MCP gateway + 34-tool Classic compatibility server
+│   ├── mcp_server.py        # Smart MCP gateway + 35-tool Classic compatibility server
 │   ├── dashboard_app.py     # dashboard WebUI (FastAPI)
 │   ├── dashboard_assets/    # primary Ledger interface + graph engine
 │   ├── classic_assets/      # selectable full operator dashboard backup

docs/AGENT_CONNECT.md

@@ -38,7 +38,7 @@ middleware. Do not expose it through a LAN address or proxy. For a remote deploy
 `engraphis[all]`, set a strong `ENGRAPHIS_API_TOKEN`, terminate TLS, and use the dashboard's
 authenticated `/mcp` endpoint instead.
 
-Use `engraphis-mcp-http --classic` only for an existing integration that requires the 34 direct
+Use `engraphis-mcp-http --classic` only for an existing integration that requires the 35 direct
 tool names. New integrations should keep the nine-tool Smart default.
 
 Engraphis documents and tests generic MCP transports; it does not claim client-specific support

docs/KILO_CODE_INTEGRATION.md

@@ -223,7 +223,7 @@ class, and the appropriate executor revalidates all of it before running.
 | `engraphis_conflict_review` | List pending/quarantined/conflicted records for review (read-only inbox). |
 
 `engraphis-mcp-classic` is only for an existing configuration that pins direct tool names. It
-preserves the former 34-tool surface below; new Kilo Code installations should keep the zero-config
+preserves the former 35-tool surface below; new Kilo Code installations should keep the zero-config
 Smart command shown above.
 
 ### Classic 35-tool inventory

engraphis/classic_assets/index.html

@@ -350,6 +350,6 @@
      graph view. dashboard.js fetches both on demand from graphRender(); see loadForceGraph()
      and loadGraphEngine(). scripts/externalize_dashboard_assets.py enforces both halves:
      they stay out of this file, and the lazy references still have to resolve. -->
-<script src="/classic-assets/dashboard.js?v=20260814-all-controls-2"></script>
+<script src="/classic-assets/dashboard.js?v=20260819-tuned-physics-final"></script>
 </body>
 </html>

engraphis/cloud_features.py

@@ -635,6 +635,10 @@ def run_managed_job(service: Any, workspace: str, kind: str, *,
     if not resolved_id:
         raise CloudFeatureError("The selected workspace does not exist.", status=404)
     cloud = client or CloudFeatureClient.from_environment(resolved_id)
+    # Read the entitlement-gated policy before taking the local write lock. The snapshot
+    # builder reserves and commits a generation, so uploading first would let a lapsed
+    # subscriber mutate local state before the Cloud control plane returns 402.
+    cloud.get_policy(resolved_id)
     workspace_id, snapshot = build_managed_snapshot(service, workspace)
     receipt = cloud.upload_snapshot(workspace_id, snapshot)
     generation = int(receipt.get("generation", snapshot["generation"]))

engraphis/core/context.py

@@ -109,15 +109,35 @@ def pack(
                 continue
 
             prefix = "\n\n" if context else ""
-            header = self._header(candidate, len(packed) + 1)
+            ordinal = len(packed) + 1
+            header = self._header(candidate, ordinal)
             base = f"{context}{prefix}{header}\n"
-            if self._count(base) >= budget:
-                continue
+            excerpt = ""
+            truncated = False
+            reason = ""
+            if self._count(base) < budget:
+                available = budget - self._count(base)
+                excerpt, truncated, reason = self._excerpt(
+                    query, candidate, available
+                )
 
-            available = budget - self._count(base)
-            excerpt, truncated, reason = self._excerpt(
-                query, candidate, available
-            )
+            # Keep the established single-pass behavior for ordinary sources.
+            # Only retry against the cheaper ordinal-only header when the selected
+            # excerpt already starts with the exact displayed title (or the titled
+            # header left no room). This removes prompt duplication without deleting
+            # evidence or weakening the stable ``[n]`` citation bridge.
+            if not excerpt or _starts_with_title(excerpt, record.title):
+                compact_base = (
+                    f"{context}{prefix}"
+                    f"{self._header(candidate, ordinal, include_title=False)}\n"
+                )
+                if self._count(compact_base) < budget:
+                    compact_available = budget - self._count(compact_base)
+                    compact = self._excerpt(query, candidate, compact_available)
+                    if compact[0] and _starts_with_title(compact[0], record.title):
+                        base = compact_base
+                        available = compact_available
+                        excerpt, truncated, reason = compact
             if not excerpt:
                 continue
             proposed = f"{base}{excerpt}"
@@ -370,15 +390,21 @@ def semantically_safe(excerpt: str) -> bool:
                 high = middle - 1
         return best
 
-    def _header(self, candidate: Candidate, ordinal: int) -> str:
+    def _header(
+        self,
+        candidate: Candidate,
+        ordinal: int,
+        *,
+        include_title: bool = True,
+    ) -> str:
         record = candidate.record
         if record is None:
             return f"[{ordinal}]"
         # The compact source list carries identity/scope. Repeating ULIDs and
         # scope labels inside the context spends reader tokens without adding
         # evidence; the ordinal is the citation bridge.
         header = f"[{ordinal}]"
-        if record.title:
+        if include_title and record.title:
             title = " ".join(record.title.split())[:120]
             header += f" {title}"
         return header
@@ -415,6 +441,18 @@ def _terms(text: str) -> set[str]:
     return {match.group(0).casefold() for match in _WORD_RE.finditer(text or "")}
 
 
+def _starts_with_title(excerpt: str, title: str) -> bool:
+    """Whether an excerpt already opens with the exact displayed title text."""
+    displayed_title = " ".join((title or "").split())[:120].casefold()
+    normalized_excerpt = " ".join((excerpt or "").split()).casefold()
+    if not displayed_title or not normalized_excerpt.startswith(displayed_title):
+        return False
+    return (
+        len(normalized_excerpt) == len(displayed_title)
+        or not normalized_excerpt[len(displayed_title)].isalnum()
+    )
+
+
 def _family_representatives(
     candidates: list[Candidate],
 ) -> tuple[list[Candidate], int]: