superlocalmemory 4.1.8 → 4.1.11

package/.claude-plugin/marketplace.json

@@ -21,7 +21,7 @@
       "license": "AGPL-3.0-or-later",
       "name": "superlocalmemory",
       "source": "./plugin",
-      "version": "4.1.8"
+      "version": "4.1.11"
     },
     {
       "author": {
@@ -39,7 +39,7 @@
       "license": "AGPL-3.0-or-later",
       "name": "superlocalmemory-codex",
       "source": "./codex-plugin",
-      "version": "4.1.8"
+      "version": "4.1.11"
     }
   ]
 }

package/CHANGELOG.md

@@ -5,6 +5,41 @@ All notable changes to SuperLocalMemory will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [4.1.11] — MCP loops use the owned runtime
+
+### Fixed
+- MCP bounded loops now route both recall gates and ledger writes through the
+  daemon, and normalize ledger tags at the HTTP boundary.
+
+## [4.1.10] — MCP loops use the owned writer
+
+### Fixed
+- MCP bounded loops now persist their ledger through the daemon/worker writer
+  instead of calling a write method on the intentionally LIGHT MCP engine, and
+  reject stale cross-profile writes rather than misplacing audit records.
+- The benchmark bundle now runs experiments 12 and 13; the emergency-recall
+  fixture no longer expires as the calendar advances, and trial totals reflect
+  every executed ablation round.
+
+## [4.1.9] — Silence is not a verdict
+
+### Fixed
+- **A recall nobody engaged with was scored as an average one.** How useful a recall
+  was is worked out from what happened next — whether a memory was quoted, edited, or
+  asked about again. When nothing at all had happened, that calculation still landed
+  on a middling score, and it was saved as if it had been measured. That is not a
+  blank: downstream it reads as mild approval, nudging results toward whatever was
+  shown and making the system steadily more certain it has no preference either way.
+  Recalls with nothing observed are now closed without a score, which is what the
+  batch pass already did — the two paths simply disagreed. On a store in daily use,
+  every one of 492 saved scores turned out to be this value, so nothing recorded there
+  reflected real use.
+- **Signals recorded as explicitly absent were counted as present.** A recall whose
+  signals were all recorded as "did not happen" was treated as having something to
+  say, and scored, because the record existed even though everything in it was empty.
+  Both paths now ask whether anything actually happened rather than whether a record
+  was written.
+
 ## [4.1.8] — The same question, answered the same way
 
 ### Fixed

package/README.md

@@ -5,7 +5,7 @@
   </picture>
 </p>
 
-<h1 align="center">SuperLocalMemory V4.1.8</h1>
+<h1 align="center">SuperLocalMemory V4.1.11</h1>
 
 <h2 align="center">Rent the LLM. Own the memory.</h2>
 
@@ -27,12 +27,12 @@ guarantee here is stated as a falsifiable invariant, tested under an adversarial
 negative control, and shipped with the harness that regenerates the evidence:
 <code>python benchmark/run_all.py --trials 200 --output-dir results/</code>. What each experiment
 does <em>not</em> exercise is stated too.</p>
-<p align="center"><code>v4.1.8</code> — one control plane: <strong>SLM-Mesh</strong> peer coordination · multi-scope memory (personal / shared / global) · profiles · Cache · Compress · 7-layer retrieval · code graph · Entity Explorer · skill evolution · Modes A/B/C · GDPR retention &amp; audit chain · bounded loops — across CLI, MCP, dashboard, the <strong>Claude plugin</strong>, the <strong>Codex add-on</strong>, and documented IDE integrations.<br/>
+<p align="center"><code>v4.1.11</code> — one control plane: <strong>SLM-Mesh</strong> peer coordination · multi-scope memory (personal / shared / global) · profiles · Cache · Compress · 7-layer retrieval · code graph · Entity Explorer · skill evolution · Modes A/B/C · GDPR retention &amp; audit chain · bounded loops — across CLI, MCP, dashboard, the <strong>Claude plugin</strong>, the <strong>Codex add-on</strong>, and documented IDE integrations.<br/>
 Proxy: <code>slm wrap claude</code> &nbsp;·&nbsp; MCP: add <code>slm_compress</code> to your config &nbsp;·&nbsp; Skill: zero-config</p>
 <p align="center"><strong>Four public arXiv preprints</strong> · V4: <a href="https://arxiv.org/abs/2608.08253">arXiv:2608.08253</a> · companion archive: <a href="https://zenodo.org/records/21853302">Zenodo 21853302</a> (<a href="https://doi.org/10.5281/zenodo.21853302">DOI 10.5281/zenodo.21853302</a>) · prior preprints: <a href="https://arxiv.org/abs/2603.02240">2603.02240</a> · <a href="https://arxiv.org/abs/2603.14588">2603.14588</a> · <a href="https://arxiv.org/abs/2604.04514">2604.04514</a>.</p>
 
 <p align="center">
-  <a href="CHANGELOG.md"><img src="https://img.shields.io/badge/v4.1.8-Current_Release-2ea44f?style=for-the-badge&logo=checkmarx&logoColor=white" alt="v4.1.8 — Current Release"/></a>
+  <a href="CHANGELOG.md"><img src="https://img.shields.io/badge/v4.1.11-Current_Release-2ea44f?style=for-the-badge&logo=checkmarx&logoColor=white" alt="v4.1.11 — Current Release"/></a>
   <a href="https://arxiv.org/abs/2608.08253"><img src="https://img.shields.io/badge/arXiv-2608.08253-b31b1b?style=for-the-badge&logo=arxiv&logoColor=white" alt="SuperLocalMemory 4.0 paper on arXiv:2608.08253"/></a>
   <a href="https://zenodo.org/records/21853302"><img src="https://img.shields.io/badge/Zenodo-10.5281%2Fzenodo.21853302-1682D4?style=for-the-badge&logo=zenodo&logoColor=white" alt="V4 paper on Zenodo: 10.5281/zenodo.21853302"/></a>
   <a href="https://arxiv.org/abs/2603.14588"><img src="https://img.shields.io/badge/arXiv-2603.14588-b31b1b?style=for-the-badge&logo=arxiv&logoColor=white" alt="arXiv Paper"/></a>

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "superlocalmemory",
-  "version": "4.1.8",
+  "version": "4.1.11",
   "description": "Local-first agent memory with MCP and an agent-native CLI. Documented clients include Claude Code, Cursor, and Windsurf.",
   "keywords": [
     "ai-memory",

package/plugin/.claude-plugin/plugin.json

@@ -15,5 +15,5 @@
   "mcpServers": "./.mcp.json",
   "name": "superlocalmemory",
   "repository": "https://github.com/qualixar/superlocalmemory",
-  "version": "4.1.8"
+  "version": "4.1.11"
 }

package/plugin/CLAUDE.md

@@ -1,4 +1,4 @@
-<!-- BEGIN SuperLocalMemory v4.1.8 -->
+<!-- BEGIN SuperLocalMemory v4.1.11 -->
 
 ## SuperLocalMemory (SLM) — Agent Rules
 
@@ -39,6 +39,6 @@ slm-recall · slm-remember · slm-session · slm-status · slm-cache · slm-comp
 ### Subagents
 slm-memory-advisor (memory decisions, session hygiene, scope/profile guidance) · slm-optimize-advisor (context compression + KV cache) · slm-governance-advisor (scope/roles/compliance/GDPR)
 
-<!-- END SuperLocalMemory v4.1.8 -->
+<!-- END SuperLocalMemory v4.1.11 -->
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/agents/slm-governance-advisor.md

@@ -77,4 +77,4 @@ slm-scope · slm-governance · slm-profile · slm-remember · slm-recall
 # What NOT to do
 Never session_init twice; never forget without dry-run preview; never store secrets; never bypass role checks; never claim an erasure succeeded without verifying via recall.
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/agents/slm-loop-runner.md

@@ -68,4 +68,4 @@ assessment. The gate is the authority.
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/agents/slm-memory-advisor.md

@@ -46,4 +46,4 @@ slm-recall · slm-remember · slm-session · slm-scope · slm-profile · slm-gov
 # What NOT to do
 Never session_init twice; never forget dry_run=False without reporting preview; never dump a whole file into remember; never invent a memory; never claim "saved" without success:true / clean CLI exit; never bypass scope or governance restrictions.
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/agents/slm-optimize-advisor.md

@@ -41,4 +41,4 @@ slm-compress · slm-cache · slm-status · slm-profile
 # What NOT to do
 Never compress code-for-edit/JSON-to-parse/<500 chars; never store secrets/ccr_ids; never let optimize failure block/alter the task; never claim a specific savings %; never carry ccr_ids across profile switches.
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/requirements.txt

@@ -1 +1 @@
-superlocalmemory==4.1.8
+superlocalmemory==4.1.11

package/plugin/skills/slm-cache/SKILL.md

@@ -145,4 +145,4 @@ These subcommands control daemon-level cache settings. They do not read or write
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/skills/slm-compress/SKILL.md

@@ -147,4 +147,4 @@ Content over 1 MB (1 000 000 bytes UTF-8) is processed but `reversible` is force
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/skills/slm-governance/SKILL.md

@@ -245,4 +245,4 @@ Before running any destructive operation (`forget`, `compact_memories`):
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin/skills/slm-graph/SKILL.md

@@ -312,4 +312,4 @@ profile. See `slm-profile` for the full profile switching workflow.
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/skills/slm-loop/SKILL.md

@@ -96,4 +96,4 @@ paused, name the approval needed; when errored, quote the short detail.
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin/skills/slm-mesh/SKILL.md

@@ -279,4 +279,4 @@ mesh availability.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin/skills/slm-profile/SKILL.md

@@ -146,4 +146,4 @@ Name them differently in your MCP config (e.g. `superlocalmemory-personal` and
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin/skills/slm-recall/SKILL.md

@@ -323,4 +323,4 @@ before recalling, then switch back. See `slm-profile` for workspace switching.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin/skills/slm-remember/SKILL.md

@@ -270,4 +270,4 @@ different workspace, use `switch_profile` first. See `slm-profile`.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin/skills/slm-scope/SKILL.md

@@ -173,4 +173,4 @@ to review the impact. See `slm-remember` for the full deletion discipline.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin/skills/slm-session/SKILL.md

@@ -253,4 +253,4 @@ explicitly and call `recall` with `include_global`/`include_shared` after
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin/skills/slm-status/SKILL.md

@@ -163,4 +163,4 @@ multi-profile setup. To switch the active profile, see `slm-profile`.
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/agents/slm-memory-advisor.md

@@ -46,4 +46,4 @@ slm-recall · slm-remember · slm-session · slm-scope · slm-profile · slm-gov
 # What NOT to do
 Never session_init twice; never forget dry_run=False without reporting preview; never dump a whole file into remember; never invent a memory; never claim "saved" without success:true / clean CLI exit; never bypass scope or governance restrictions.
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/agents/slm-optimize-advisor.md

@@ -41,4 +41,4 @@ slm-compress · slm-cache · slm-status · slm-profile
 # What NOT to do
 Never compress code-for-edit/JSON-to-parse/<500 chars; never store secrets/ccr_ids; never let optimize failure block/alter the task; never claim a specific savings %; never carry ccr_ids across profile switches.
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/rules/AGENTS.md

@@ -137,4 +137,4 @@ When the SLM MCP server is unavailable, use these CLI equivalents:
 - **slm-optimize-advisor** — context compression and KV cache
 - **slm-governance-advisor** — scope/role compliance, retention policies, GDPR
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/skills/slm-cache/SKILL.md

@@ -145,4 +145,4 @@ These subcommands control daemon-level cache settings. They do not read or write
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/skills/slm-compress/SKILL.md

@@ -147,4 +147,4 @@ Content over 1 MB (1 000 000 bytes UTF-8) is processed but `reversible` is force
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/skills/slm-governance/SKILL.md

@@ -245,4 +245,4 @@ Before running any destructive operation (`forget`, `compact_memories`):
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin-src/skills/slm-graph/SKILL.md

@@ -312,4 +312,4 @@ profile. See `slm-profile` for the full profile switching workflow.
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/skills/slm-loop/SKILL.md

@@ -96,4 +96,4 @@ paused, name the approval needed; when errored, quote the short detail.
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/plugin-src/skills/slm-mesh/SKILL.md

@@ -279,4 +279,4 @@ mesh availability.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin-src/skills/slm-profile/SKILL.md

@@ -146,4 +146,4 @@ Name them differently in your MCP config (e.g. `superlocalmemory-personal` and
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin-src/skills/slm-recall/SKILL.md

@@ -323,4 +323,4 @@ before recalling, then switch back. See `slm-profile` for workspace switching.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin-src/skills/slm-remember/SKILL.md

@@ -270,4 +270,4 @@ different workspace, use `switch_profile` first. See `slm-profile`.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin-src/skills/slm-scope/SKILL.md

@@ -173,4 +173,4 @@ to review the impact. See `slm-remember` for the full deletion discipline.
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin-src/skills/slm-session/SKILL.md

@@ -253,4 +253,4 @@ explicitly and call `recall` with `include_global`/`include_shared` after
 
 ---
 
-*SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later*

package/plugin-src/skills/slm-status/SKILL.md

@@ -163,4 +163,4 @@ multi-profile setup. To switch the active profile, see `slm-profile`.
 
 ---
 
-SuperLocalMemory v4.1.8 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.1.11 · Qualixar · AGPL-3.0-or-later

package/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "superlocalmemory"
-version = "4.1.8"
+version = "4.1.11"
 description = "Local-first agent memory with auditable hybrid retrieval"
 readme = "README.md"
 license = "AGPL-3.0-or-later"

package/src/superlocalmemory/__init__.py

@@ -32,7 +32,7 @@ if "OMP_NUM_THREADS" not in os.environ:
     os.environ["OMP_NUM_THREADS"] = "2"
 # ---------------------------------------------------------------------------
 
-__version__ = "4.1.8"
+__version__ = "4.1.11"
 
 _REQUIRED_VERSIONS = {
     "sentence_transformers": "5.3.0",

package/src/superlocalmemory/cli/daemon.py

@@ -364,6 +364,14 @@ class DaemonRefused(RuntimeError):
         )
 
 
+class DaemonConflict(RuntimeError):
+    """A deterministic daemon conflict that the caller must resolve."""
+
+    def __init__(self, detail: str) -> None:
+        self.detail = detail or "daemon request conflicted with current state"
+        super().__init__(self.detail)
+
+
 def daemon_request(
     method: str,
     path: str,
@@ -373,6 +381,7 @@ def daemon_request(
     expected_descriptor=_EXPECTED_DESCRIPTOR_UNSET,
     expected_legacy: dict | None = None,
     verify_health: bool = True,
+    preserve_conflict: bool = False,
 ) -> dict | None:
     """Send a request only after validating the owned daemon identity.
 
@@ -463,6 +472,15 @@ def daemon_request(
         # locally as the machine owner.
         if exc.code in (401, 403):
             raise DaemonRefused(exc.code, path) from exc
+        if exc.code == 409 and preserve_conflict:
+            detail = "daemon request conflicted with current state"
+            try:
+                payload = json.loads(exc.read().decode())
+                if isinstance(payload, dict) and payload.get("detail"):
+                    detail = str(payload["detail"])
+            except Exception:
+                pass
+            raise DaemonConflict(detail) from exc
         return None
     except Exception:
         return None

package/src/superlocalmemory/learning/reward.py

@@ -113,6 +113,30 @@ _MAX_CLOCK_SKEW_MS: Final[int] = 60_000
 # ---------------------------------------------------------------------------
 
 
+#: Dwell below this is not engagement; the label formula uses the same bound.
+_DWELL_THRESHOLD_MS: Final[int] = 2000
+
+
+def _carries_evidence(signals: Mapping[str, object]) -> bool:
+    """Did anything actually happen after this recall?
+
+    ``_compute_label`` is built as ``0.5 + bonuses - penalties``, so a signal
+    set with nothing in it evaluates to exactly one half. That number is not an
+    absence of judgement, it is a confident one: it becomes a mid-strength
+    positive training label and a Beta update that tightens a posterior around
+    its prior. An empty dict and a dict of falsy values mean the same thing here
+    and must both be refused.
+    """
+    if not signals:
+        return False
+    if signals.get("cite") or signals.get("edit") or signals.get("requery"):
+        return True
+    try:
+        return int(signals.get("dwell_ms", 0) or 0) >= _DWELL_THRESHOLD_MS
+    except (TypeError, ValueError):
+        return False
+
+
 def _compute_label(signals: Mapping[str, object]) -> float:
     """Deterministic label in ``[0.0, 1.0]`` per the manifest A.1 formula.
 
@@ -141,8 +165,9 @@ def _compute_label(signals: Mapping[str, object]) -> float:
         dwell_int = 0
 
     dwell_bonus = 0.0
-    if dwell_int >= 2000:
-        dwell_bonus = min(0.15, 0.05 + (dwell_int - 2000) / 80_000.0)
+    if dwell_int >= _DWELL_THRESHOLD_MS:
+        dwell_bonus = min(
+            0.15, 0.05 + (dwell_int - _DWELL_THRESHOLD_MS) / 80_000.0)
 
     label = (
         0.5
@@ -585,9 +610,19 @@ class EngagementRewardModel:
                     signals = json.loads(pending["signals_json"] or "{}")
                 except json.JSONDecodeError:  # pragma: no cover — defensive
                     signals = {}
-                reward = _compute_label(signals)
                 now_ms = self._clock_ms()
                 timestamp_iso = _iso_from_ms(now_ms)
+                if not _carries_evidence(signals):
+                    # Finalize so it stops being rescanned, but record no
+                    # outcome: nothing was observed, and a score here would
+                    # be invented rather than measured. Mirrors reap_stale.
+                    conn.execute(
+                        "UPDATE pending_outcomes "
+                        "SET status = 'settled' WHERE outcome_id = ?",
+                        (outcome_id,),
+                    )
+                    return _FALLBACK_REWARD
+                reward = _compute_label(signals)
 
                 # NOTE: Split INSERT across lines so the Stage-5b CI
                 # gate's single-line regex (LLD-00 §13) does not fire.
@@ -746,7 +781,7 @@ class EngagementRewardModel:
             #
             # The row is still finalized, so it stops being rescanned; it is
             # simply not turned into evidence it never was.
-            if not signals:
+            if not _carries_evidence(signals):
                 settle_ids.append(row["outcome_id"])
                 continue
             reward = _compute_label(signals)

package/src/superlocalmemory/loops/__init__.py

@@ -21,6 +21,7 @@ from superlocalmemory.loops.ledger import (
     SLMMemoryLedger,
     engine_backed_ledger,
     open_engine_store,
+    pool_backed_ledger,
 )
 from superlocalmemory.loops.models import (
     Bounds,
@@ -49,6 +50,7 @@ __all__ = [
     "InMemoryLedger",
     "SLMMemoryLedger",
     "engine_backed_ledger",
+    "pool_backed_ledger",
     "open_engine_store",
     "no_progress",
     "rung_requires_approval",

package/src/superlocalmemory/loops/ledger.py

@@ -223,6 +223,36 @@ class _EngineLedgerStore:
             self._engine.close()
 
 
+class _PoolLedgerStore(_EngineLedgerStore):
+    """Write through the owned daemon/worker; read through the LIGHT engine.
+
+    MCP intentionally keeps a LIGHT engine in-process.  That engine owns a
+    read-only database view but refuses ``store``/``store_fast``.  The pool is
+    the admitted writer and shares the same canonical store, so loop laps stay
+    immediately queryable without turning every MCP process into a heavy
+    writer.
+    """
+
+    def __init__(self, pool: Any, reader_engine: Any) -> None:
+        super().__init__(reader_engine, owns_engine=False)
+        self._pool = pool
+
+    def add(self, content: str, *, session_id: str, metadata: dict) -> None:
+        result = self._pool.store(
+            content,
+            metadata={
+                **metadata,
+                "session_id": session_id,
+                "profile_id": self._engine.profile_id,
+            },
+        )
+        accepted = bool(result.get("ok") or result.get("success"))
+        if not accepted:
+            raise RuntimeError(
+                result.get("error", "owned SLM writer rejected loop ledger entry")
+            )
+
+
 def engine_backed_ledger(engine: Any) -> SLMMemoryLedger:
     """Build an SLM-backed ledger over an ALREADY-OPEN engine.
 
@@ -236,6 +266,11 @@ def engine_backed_ledger(engine: Any) -> SLMMemoryLedger:
     return SLMMemoryLedger(_EngineLedgerStore(engine, owns_engine=False))
 
 
+def pool_backed_ledger(pool: Any, reader_engine: Any) -> SLMMemoryLedger:
+    """Build an MCP-safe ledger: admitted pool writes, LIGHT engine reads."""
+    return SLMMemoryLedger(_PoolLedgerStore(pool, reader_engine))
+
+
 def open_engine_store(db_path: str | Path) -> _EngineLedgerStore:
     """Build an engine-backed ledger store rooted at ``db_path``.
 

package/src/superlocalmemory/mcp/_daemon_proxy.py

@@ -139,21 +139,40 @@ class DaemonPoolProxy:
     ) -> dict[str, Any]:
         if self._unavailable:
             return self._unavailable_response()
+        tags = (metadata or {}).get("tags", "")
+        if isinstance(tags, (list, tuple, set)):
+            tags = ",".join(str(tag) for tag in tags)
         body = {
             "content": content,
-            "tags": (metadata or {}).get("tags", ""),
+            "tags": tags,
             "metadata": metadata or {},
             "session_id": (metadata or {}).get("session_id", ""),
             "idempotency_key": (metadata or {}).get("idempotency_key") or None,
+            "profile_id": (metadata or {}).get("profile_id", ""),
         }
+        # One identity-aware daemon client owns descriptor validation,
+        # capability delivery, and exact-instance targeting. A raw urllib POST
+        # here previously became unauthenticated when /remember was hardened
+        # and could also attach to a stale/foreign port.
         try:
-            # One identity-aware daemon client owns descriptor validation,
-            # capability delivery, and exact-instance targeting.  A raw urllib
-            # POST here previously became unauthenticated when /remember was
-            # hardened and could also attach to a stale/foreign port.
-            from superlocalmemory.cli.daemon import daemon_request
-
-            data = daemon_request("POST", "/remember", body)
+            from superlocalmemory.cli.daemon import DaemonConflict, daemon_request
+        except Exception as exc:
+            logger.warning("daemon client import failed: %s", exc)
+            return self._unavailable_response()
+        try:
+            data = daemon_request(
+                "POST",
+                "/remember",
+                body,
+                preserve_conflict=True,
+            )
+        except DaemonConflict as exc:
+            return {
+                "ok": False,
+                "code": "PROFILE_MISMATCH",
+                "retryable": False,
+                "error": str(exc),
+            }
         except Exception as exc:
             logger.warning("daemon /remember failed: %s", exc)
             return self._unavailable_response()

package/src/superlocalmemory/mcp/server.py

@@ -280,7 +280,12 @@ register_evolution_tools(_target, get_engine)  # v3.4.11: Skill evolution tools
 from superlocalmemory.mcp.tools_optimize import register_optimize_tools
 register_optimize_tools(_target)  # v3.6.11: Surface B Optimize tools (proxy-free)
 from superlocalmemory.mcp.tools_loops import register_loop_tools
-register_loop_tools(_target, get_engine)  # v3.8.0: bounded-loop tools (CLI+command+MCP)
+from superlocalmemory.mcp._daemon_proxy import choose_pool as _choose_loop_pool
+register_loop_tools(
+    _target,
+    get_engine,
+    _choose_loop_pool,
+)  # v3.8.0: bounded-loop tools (CLI+command+MCP)
 from superlocalmemory.mcp.tools_ops import register_ops_tools
 register_ops_tools(_target, get_engine)  # operational recovery & admin remediation
 from superlocalmemory.mcp.tools_brain import register_brain_tools

package/src/superlocalmemory/mcp/tools_loops.py

@@ -47,6 +47,7 @@ from superlocalmemory.loops import (
     LedgerEntry,
     Verdict,
     engine_backed_ledger,
+    pool_backed_ledger,
     run_bounded_loop,
 )
 
@@ -69,7 +70,11 @@ _MAX_NAME_CHARS = 128
 _MAX_QUERY_CHARS = 2000
 
 
-def register_loop_tools(server, get_engine: Callable) -> None:
+def register_loop_tools(
+    server,
+    get_engine: Callable,
+    get_pool: Callable | None = None,
+) -> None:
     """Register the 3 bounded-loop tools on *server*.
 
     *server* is duck-typed: must support the ``@server.tool()`` decorator.
@@ -144,10 +149,46 @@ def register_loop_tools(server, get_engine: Callable) -> None:
             min_score = float(gate_min_score)
 
             engine = get_engine()
+            pool = get_pool() if get_pool is not None else None
 
             def gate(lap: int) -> Verdict:
-                resp = engine.recall(gate_query, limit=3, fast=True)
-                all_results = getattr(resp, "results", None) or []
+                if pool is not None:
+                    resp = pool.recall(gate_query, limit=3, fast=True)
+                    if not isinstance(resp, dict) or resp.get("ok") is False:
+                        raise RuntimeError(
+                            (resp or {}).get("error", "owned SLM reader rejected recall")
+                            if isinstance(resp, dict)
+                            else "owned SLM reader returned an invalid response"
+                        )
+                    all_results = resp.get("results", []) or []
+                    floored = bool(resp.get("no_confident_match", False))
+                else:
+                    resp = engine.recall(gate_query, limit=3, fast=True)
+                    all_results = getattr(resp, "results", None) or []
+                    floored = bool(getattr(resp, "no_confident_match", False))
+
+                def _content(result: Any) -> str:
+                    if isinstance(result, dict):
+                        fact = result.get("fact")
+                        if isinstance(fact, dict):
+                            return str(fact.get("content", "") or "")
+                        return str(result.get("content", "") or "")
+                    return str(
+                        getattr(getattr(result, "fact", None), "content", "") or ""
+                    )
+
+                def _score(result: Any) -> float:
+                    if isinstance(result, dict):
+                        value = result.get("score", result.get("relevance_score", 0.0))
+                    else:
+                        value = getattr(result, "score", None)
+                        if value is None:
+                            value = getattr(result, "relevance_score", 0.0) or 0.0
+                    try:
+                        return float(value)
+                    except (TypeError, ValueError):
+                        return 0.0
+
                 # Exclude the loop's own audit records from the gate evaluation.
                 # Each lap writes a LedgerEntry (JSON with "run_id" + "lap") into
                 # SLM via store_fast; the FTS5 trigger indexes every such insert.
@@ -157,20 +198,9 @@ def register_loop_tools(server, get_engine: Callable) -> None:
                 # can satisfy an independent gate.
                 results = [
                     r for r in all_results
-                    if LedgerEntry.from_json(
-                        getattr(getattr(r, "fact", None), "content", "") or ""
-                    ) is None
+                    if LedgerEntry.from_json(_content(r)) is None
                 ]
-                floored = bool(getattr(resp, "no_confident_match", False))
-                best: float = 0.0
-                for r in results:
-                    s = getattr(r, "score", None)
-                    if s is None:
-                        s = getattr(r, "relevance_score", 0.0) or 0.0
-                    try:
-                        best = max(best, float(s))
-                    except (TypeError, ValueError):
-                        pass
+                best = max((_score(result) for result in results), default=0.0)
                 passed = bool(results) and not floored and best >= min_score
                 return Verdict(
                     passed,
@@ -187,7 +217,11 @@ def register_loop_tools(server, get_engine: Callable) -> None:
                     time.sleep(poll)
                 return LapResult(changed=False, tokens=0)
 
-            ledger = engine_backed_ledger(engine)
+            ledger = (
+                pool_backed_ledger(pool, engine)
+                if pool is not None
+                else engine_backed_ledger(engine)
+            )
 
             # The loop blocks (sleeps between laps); run it off the event loop.
             # The engine's per-call WAL connection model makes this thread-safe.

package/src/superlocalmemory/server/unified_daemon.py

@@ -630,6 +630,10 @@ class RememberRequest(BaseModel):
     metadata: dict | None = None  # v3.4.26: pass-through from MCP pool_store
     idempotency_key: str | None = None
     session_id: str = ""
+    # Optional compare-and-write guard for profile-sensitive clients. The
+    # daemon remains single-profile, so a stale MCP client must fail instead
+    # of silently writing into whichever profile another client selected.
+    profile_id: str = ""
     #: WHEN this memory is about, as distinct from when it was written.
     #:
     #: The internal admission record has carried this field all along and this
@@ -693,6 +697,18 @@ class RememberRequest(BaseModel):
     shared_with: list[str] | None = None
 
 
+def _require_remember_profile(requested: str, active: str) -> None:
+    """Reject a stale profile-bound write before any durable mutation."""
+    if requested and requested != active:
+        raise HTTPException(
+            status_code=409,
+            detail=(
+                f"profile mismatch: request is bound to '{requested}' but "
+                f"the daemon currently serves '{active}'"
+            ),
+        )
+
+
 class SessionOpenRequest(BaseModel):
     # #49: local session-open warm (no model roundtrip needed)
     project_path: str = ""
@@ -4671,6 +4687,7 @@ def _register_daemon_routes(application: FastAPI) -> None:
         trusted_actor_id = _require_write_actor(request)
         _update_activity()
         engine = _get_engine_or_503()
+        _require_remember_profile(req.profile_id, engine._profile_id)
 
         # v3.6.15 multi-scope: resolve the write scope. ``None`` (not specified
         # by the caller) → the configured default_scope (personal). Shared

package/src/superlocalmemory/ui/index.html

@@ -1587,7 +1587,7 @@
     <script src="static/js/event-delegation.js?v=379"></script>
     <!-- OD screen modules (v3.7.9): approved-design panes wired to live data.
          Load LAST so their window.load<Screen> overrides win over the legacy loaders. -->
-    <script src="static/js/od-brain.js?v=4f233c96"></script>
+    <script src="static/js/od-brain.js?v=77369272"></script>
     <script src="static/js/od-components.js?v=382"></script>
     <script src="static/js/od-health.js?v=382"></script>
     <script src="static/js/od-operations.js?v=380"></script>
@@ -1595,7 +1595,7 @@
     <script src="static/js/od-ops-health.js?v=400"></script>
     <script src="static/js/od-team.js?v=379"></script>
     <script src="static/js/od-graph.js?v=6812bf6c"></script>
-    <script src="static/js/od-memories.js?v=997f1674"></script>
+    <script src="static/js/od-memories.js?v=30ac32ed"></script>
     <script src="static/js/od-entities.js?v=379"></script>
     <!-- Multi-Agent Memory pane (v3.8.0): visualises memory written by multiple agents -->
     <script src="static/js/od-agents.js?v=c75ff3c2"></script>