superlocalmemory 4.2.2__tar.gz → 4.2.3__tar.gz

{superlocalmemory-4.2.2/src/superlocalmemory.egg-info → superlocalmemory-4.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: superlocalmemory
-Version: 4.2.2
+Version: 4.2.3
 Summary: The governed memory brain for AI agents: one local memory for Claude Code, Codex, Cursor, Copilot, ChatGPT and any MCP client
 Author-email: Varun Pratap Bhardwaj <admin@superlocalmemory.com>
 License-Expression: AGPL-3.0-or-later
@@ -148,9 +148,7 @@ In Mode A, core remember and recall make no model-provider call: no API key, no
 
 ## Install
 
-Three ways in. All three end in the same setup: your text memory on this machine, and, if you say yes, your pictures and documents too.
-
-**1. Ask your agent.** Copy this into Claude Code, Codex, Cursor or any agent that can run commands:
+> **The easiest way: let your AI agent install it.** Copy the block below into Claude Code, Codex, Cursor or any agent that can run commands. It reads the setup guide, asks you one question, sets everything up and tells you what works. No terminal commands for you to learn.
 
 ```
 Install SuperLocalMemory on this computer for me. Follow
@@ -161,7 +159,17 @@ screenshots and PDFs by describing them. Connect SuperLocalMemory to the agent
 I am using. When you finish, run slm doctor and tell me in plain words what works.
 ```
 
-**2. npm**, the primary route (Node 18+). It installs SLM into a package-owned virtual environment:
+**No AI agent? One line in Terminal** (macOS and Linux). It brings its own Python, sets everything up, connects the AI tools it finds and opens the dashboard; no Python or npm needed beforehand:
+
+```bash
+curl -fsSLo get-slm.sh https://raw.githubusercontent.com/qualixar/superlocalmemory/main/scripts/get-slm.sh && sh get-slm.sh
+```
+
+Add pictures and documents from the start with `sh get-slm.sh --media` at the end. It downloads the script first, so you can read it before it runs. Windows: use the agent prompt above, or npm or pip below.
+
+**Prefer the terminal?** Both routes end in the same setup: your text memory on this machine, and, if you say yes, your pictures and documents too.
+
+**npm**, the primary route (Node 18+). It installs SLM into a package-owned virtual environment:
 
 ```bash
 npm install -g superlocalmemory   # add --media to also turn on pictures and documents
@@ -169,7 +177,7 @@ slm setup                         # pick Mode A to keep everything on this machi
 slm connect claude-code           # or codex, cursor, windsurf, zed, claude-desktop, jetbrains, opencode ...
 ```
 
-**3. pip**, the other primary route (Python 3.12 to 3.14), in a Python virtual environment you activate:
+**pip**, the other primary route (Python 3.12 to 3.14), in a Python virtual environment you activate:
 
 ```bash
 python3 -m venv .venv

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/README.md

@@ -32,9 +32,7 @@ In Mode A, core remember and recall make no model-provider call: no API key, no
 
 ## Install
 
-Three ways in. All three end in the same setup: your text memory on this machine, and, if you say yes, your pictures and documents too.
-
-**1. Ask your agent.** Copy this into Claude Code, Codex, Cursor or any agent that can run commands:
+> **The easiest way: let your AI agent install it.** Copy the block below into Claude Code, Codex, Cursor or any agent that can run commands. It reads the setup guide, asks you one question, sets everything up and tells you what works. No terminal commands for you to learn.
 
 ```
 Install SuperLocalMemory on this computer for me. Follow
@@ -45,7 +43,17 @@ screenshots and PDFs by describing them. Connect SuperLocalMemory to the agent
 I am using. When you finish, run slm doctor and tell me in plain words what works.
 ```
 
-**2. npm**, the primary route (Node 18+). It installs SLM into a package-owned virtual environment:
+**No AI agent? One line in Terminal** (macOS and Linux). It brings its own Python, sets everything up, connects the AI tools it finds and opens the dashboard; no Python or npm needed beforehand:
+
+```bash
+curl -fsSLo get-slm.sh https://raw.githubusercontent.com/qualixar/superlocalmemory/main/scripts/get-slm.sh && sh get-slm.sh
+```
+
+Add pictures and documents from the start with `sh get-slm.sh --media` at the end. It downloads the script first, so you can read it before it runs. Windows: use the agent prompt above, or npm or pip below.
+
+**Prefer the terminal?** Both routes end in the same setup: your text memory on this machine, and, if you say yes, your pictures and documents too.
+
+**npm**, the primary route (Node 18+). It installs SLM into a package-owned virtual environment:
 
 ```bash
 npm install -g superlocalmemory   # add --media to also turn on pictures and documents
@@ -53,7 +61,7 @@ slm setup                         # pick Mode A to keep everything on this machi
 slm connect claude-code           # or codex, cursor, windsurf, zed, claude-desktop, jetbrains, opencode ...
 ```
 
-**3. pip**, the other primary route (Python 3.12 to 3.14), in a Python virtual environment you activate:
+**pip**, the other primary route (Python 3.12 to 3.14), in a Python virtual environment you activate:
 
 ```bash
 python3 -m venv .venv

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/agents/slm-governance-advisor.md

@@ -81,4 +81,4 @@ slm-scope · slm-governance · slm-profile · slm-remember · slm-recall · slm-
 # What NOT to do
 Never session_init twice; never forget without dry-run preview; never run `slm gdpr erase --yes` unasked; never store secrets; never bypass role checks; never claim an erasure succeeded without verifying via recall.
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/agents/slm-loop-runner.md

@@ -79,4 +79,4 @@ assessment. The gate is the authority.
 
 ---
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/agents/slm-memory-advisor.md

@@ -51,4 +51,4 @@ 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 switch the active profile unasked; 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 v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/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 hand a ccr_id to another agent.
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/rules/AGENTS.md

@@ -187,4 +187,4 @@ Using this memory from a web assistant or another computer: see the slm-web-acce
 - **slm-governance-advisor** — scope/role compliance, retention and lifecycle, GDPR
 - **slm-loop-runner** — bounded, gate-verified loops (the `/slm-loop` command delegates to it)
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/rules/CLAUDE.md.fragment

@@ -1,4 +1,4 @@
-<!-- BEGIN SuperLocalMemory v4.2.2 -->
+<!-- BEGIN SuperLocalMemory v4.2.3 -->
 
 ## SuperLocalMemory (SLM) — Agent Rules
 
@@ -51,6 +51,6 @@ 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) · slm-loop-runner (bounded, gate-verified loops)
 
-<!-- END SuperLocalMemory v4.2.2 -->
+<!-- END SuperLocalMemory v4.2.3 -->
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-bot-memory/SKILL.md

@@ -178,4 +178,4 @@ store every local bot on that computer can reach.
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-cache/SKILL.md

@@ -144,4 +144,4 @@ These subcommands control daemon-level cache settings and bulk removal. They do
 
 ---
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/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 v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-getting-started-bot/SKILL.md

@@ -194,4 +194,4 @@ quality on a shared box.
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-governance/SKILL.md

@@ -274,4 +274,4 @@ Before running any destructive or state-changing operation (`forget`,
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-graph/SKILL.md

@@ -324,4 +324,4 @@ All are host-only: a remote caller cannot use them.
 
 ---
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-loop/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: slm-loop
-description: Gate-verified bounded loops with SuperLocalMemory as the durable ledger. Use when a task has a checkable acceptance condition and you must iterate until an INDEPENDENT gate passes — never stopping because the agent believes it is done. `slm_loop_run` (MCP) waits, under hard bounds, for a recall gate to pass; `slm loop demo` shows the control flow keyless; `slm loop history` and `slm loop show <run_id>` (or `slm_loop_history` / `slm_loop_show`) inspect past runs, whose every lap is stored as queryable SLM memory (tag `loop:<name>`). Terminal statuses are DONE / HALT / PAUSE / KILLED / ERROR — report them exactly, never converting HALT/PAUSE/ERROR into success.
+description: Gate-verified bounded loops with SuperLocalMemory as the durable ledger. Use when a task has a checkable acceptance condition and you must iterate until an INDEPENDENT gate passes — never stopping because the agent believes it is done. `slm_loop_run` (MCP) waits, under hard bounds, for an independently executed check registered by the operator to pass; `slm loop demo` shows the control flow keyless; `slm loop history` and `slm loop show <run_id>` (or `slm_loop_history` / `slm_loop_show`) inspect past runs, whose every lap is stored as queryable SLM memory (tag `loop:<name>`). Terminal statuses are DONE / HALT / PAUSE / KILLED / ERROR — report them exactly, never converting HALT/PAUSE/ERROR into success.
 when_to_use: "run a bounded loop, gate-verified task, iterate until tests pass, verify against an independent gate, don't trust the agent's own done claim, slm loop, convergence loop, loop until green, loop ledger, wait for another agent's result"
 allowed-tools: slm_loop_run, slm_loop_history, slm_loop_show, recall, report_outcome, get_soft_prompts, skill_health, skill_lineage, evolve_skill, observe_bounded_loop_evidence, Bash
 ---
@@ -26,7 +26,7 @@ subprocess or sandbox machinery. What each surface gives you:
 
 | Surface | What it does |
 |---|---|
-| `slm_loop_run` (MCP) | A *watcher*: each lap is a recall of `gate_query`; the loop ends when a confident match appears, or a bound trips. Use it to wait for a verification or coordination memory another agent will write (for example "build passed"). |
+| `slm_loop_run` (MCP) | A *watcher* for a registered check. DONE needs `gate_contract={"job_id": ...}`: a job the operator registered with `superlocalmemory.loops.receipt_jobs.register_job` and executed with `run_registered_check`. Each lap's recall of `gate_query` only locates context; a matching memory alone never passes. Without a registered job the run ends HALT ("operator-registered job_id required"), which is the correct, honest outcome. |
 | `slm loop demo` (CLI) | A keyless convergence demo: a stub proposer, a deterministic gate that passes on lap 3, every lap recorded. Proves the engine and ledger work end to end. |
 | `slm loop history`, `slm loop show` / `slm_loop_history`, `slm_loop_show` | Read the ledger of past runs. |
 | `superlocalmemory.loops.run_bounded_loop` (Python) | The engine itself, for code that supplies its own runner and gate callables. |
@@ -62,6 +62,7 @@ slm_loop_run(
   max_tokens: int = 0,          # optional token budget; 0 disables
   no_progress_window: int = 0,  # halt after N no-change laps; leave 0 for a watcher
   require_support: bool = False,# pass only if the answer check ran and judged it sufficient
+  gate_contract: dict | None = None,  # {"job_id": <operator registration>}; required for DONE
 )
 ```
 
@@ -170,4 +171,4 @@ Guides: `docs/skill-evolution.md`, `docs/bounded-loops-bridge.md`.
 
 ---
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-media/SKILL.md

@@ -337,4 +337,4 @@ minutes (previous vectors are kept) and let them decide.
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-mesh/SKILL.md

@@ -386,4 +386,4 @@ mesh availability.
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-profile/SKILL.md

@@ -161,4 +161,4 @@ anything but `core` or `mesh`.
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-recall/SKILL.md

@@ -515,4 +515,4 @@ See `slm-profile`.
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-remember/SKILL.md

@@ -372,4 +372,4 @@ sessions on this machine, so use it only when the user asks to move. See
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-scope/SKILL.md

@@ -180,4 +180,4 @@ to review the impact. See `slm-remember` for the full deletion discipline.
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-session/SKILL.md

@@ -288,4 +288,4 @@ answered. To also surface shared or global facts, call `recall` with
 
 ---
 
-*SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later*
+*SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later*

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/plugin-src/skills/slm-status/SKILL.md

@@ -228,4 +228,4 @@ the memory-engine upgrade is pending.
 
 ---
 
-SuperLocalMemory v4.2.2 · Qualixar · AGPL-3.0-or-later
+SuperLocalMemory v4.2.3 · Qualixar · AGPL-3.0-or-later

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "superlocalmemory"
-version = "4.2.2"
+version = "4.2.3"
 description = "The governed memory brain for AI agents: one local memory for Claude Code, Codex, Cursor, Copilot, ChatGPT and any MCP client"
 readme = "README.md"
 license = "AGPL-3.0-or-later"

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/__init__.py

@@ -32,7 +32,7 @@ if "OMP_NUM_THREADS" not in os.environ:
     os.environ["OMP_NUM_THREADS"] = "2"
 # ---------------------------------------------------------------------------
 
-__version__ = "4.2.2"
+__version__ = "4.2.3"
 
 _REQUIRED_VERSIONS = {
     "sentence_transformers": "5.6.1",

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/mcp/profiles.py

@@ -119,9 +119,9 @@ _PROFILE_FULL: frozenset[str] = frozenset({
     # and a view is a convenience over recall, which core already has.
     "run_view", "manage_view",
     # prestage_context remains registered but deliberately raw-server-only.
-}) | _PROFILE_FULL_MESH | _PROFILE_KINDS | _media_tools_if_enabled()  # 57 (+5 while images are on)
+}) | _PROFILE_FULL_MESH | _PROFILE_KINDS | _media_tools_if_enabled()  # 58 (+5 while images are on)
 
-_PROFILE_POWER: frozenset[str] = _PROFILE_FULL | frozenset({  # 69
+_PROFILE_POWER: frozenset[str] = _PROFILE_FULL | frozenset({  # 70 (+5 while images are on)
     "get_version", "get_mode", "health", "consistency_check", "recall_trace",
     "get_lifecycle_status", "set_retention_policy", "compact_memories",
     "get_behavioral_patterns", "audit_trail", "quantize", "get_retention_stats",

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/mcp/server.py

@@ -466,7 +466,23 @@ def _prestage_recall(query: str, limit: int, profile_id: str, as_of: str | None
         target or getattr(engine, "profile_id", "default"),
     )
     if ctx is None:
-        return _memories_from(engine.recall(query, **kwargs))
+        from superlocalmemory.core.engine_capabilities import CapabilityError
+
+        try:
+            return _memories_from(engine.recall(query, **kwargs))
+        except CapabilityError:
+            # 4.2.3: an MCP stdio process holds a light engine that may not
+            # recall in-process, so every prestage_context said recall_error.
+            # Recall through the worker pool, as session_init does. A
+            # point-in-time view is not forwarded there: refuse, never serve now.
+            if as_of is not None:
+                raise
+            from superlocalmemory.core.answer_check_scope import context_load
+            from superlocalmemory.mcp._pool_adapter import pool_recall
+
+            with context_load():
+                return _memories_from(pool_recall(
+                    query, limit=limit, **({"profile_id": target} if target else {})))
     with visibility.use(ctx):
         return _memories_from(engine.recall(query, **kwargs))
 

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/mcp/tools_core.py

@@ -583,6 +583,7 @@ def register_core_tools(server, get_engine: Callable) -> None:
         include_related: bool = False,
         depth: int = 1,
         max_tokens: int = 0,
+        budget_s: float = 0,
     ) -> dict:
         """Search memories through hybrid retrieval, RRF fusion, and reranking.
 
@@ -594,6 +595,10 @@ def register_core_tools(server, get_engine: Callable) -> None:
         neighbours, facts about the same entities, the correction that replaced
         it). Start at 0 or 1; drill into one result with ``expand``.
         ``max_tokens`` caps the returned results; it is never exceeded.
+        ``budget_s`` (seconds, optional) is for a recall that blocks a turn,
+        such as a per-message prefetch: it answers inside that time, with
+        unchecked keyword matches when full recall cannot finish. Leave it
+        unset for a question; recall then waits for the full answer.
         If ``warming_up`` is true, tell the user SuperLocalMemory is still
         warming up and to retry shortly; do not conclude the memory is absent.
         ``code: RECALL_BUSY`` means every recall slot is serving other agents:
@@ -768,6 +773,14 @@ def register_core_tools(server, get_engine: Callable) -> None:
             # worker thread below is not it). No deadline -> nothing is sent.
             from superlocalmemory.mcp.request_deadline import remaining_budget_s
             _budget_s = remaining_budget_s()
+            # 4.2.3: a turn-blocking caller (Hermes prefetch) names its own
+            # budget; it can only shorten the relay's, never lengthen it.
+            try:
+                _own = float(budget_s or 0)
+            except (TypeError, ValueError):
+                _own = 0.0
+            if _own > 0 and _own == _own:
+                _budget_s = min(_own, _budget_s) if _budget_s is not None else _own
 
             def _recall_via_daemon_pool():
                 pool = choose_pool()
@@ -1443,6 +1456,18 @@ def register_core_tools(server, get_engine: Callable) -> None:
             )
             from superlocalmemory.learning.behavioral import BehavioralPatternStore
             store = BehavioralPatternStore(engine._db.db_path)
+            # 4.2.3: a correction must name a pattern that exists (its id or
+            # its key, as get_behavioral_patterns lists them); an unknown one
+            # used to return success and record an annotation for nothing.
+            wanted = str(pattern_id or "").strip()
+            known = store.get_patterns(pid, limit=100000)
+            if not wanted or not any(
+                    wanted in (str(p.get("id", "")), str(p.get("pattern_key", "")))
+                    for p in known):
+                authorization.complete()
+                return {"success": False, "retryable": False,
+                        "error": f"no behavioral pattern {wanted!r} in this profile; "
+                                 "use get_behavioral_patterns for ids"}
             # The store's write is record_pattern; the pattern key travels in
             # its data. (Before 4.1.21 this called a method the store does not
             # have, so every correction failed.)
@@ -1669,13 +1694,41 @@ def register_core_tools(server, get_engine: Callable) -> None:
 
             path = with_view(
                 "/api/corrections/" + urllib.parse.quote(case_id, safe="") + "/" + action)
-            result = await asyncio.to_thread(daemon_request, "POST", path, payload)
+            from superlocalmemory.cli.daemon import (
+                DaemonConflict,
+                DaemonNotFound,
+                DaemonRefused,
+                DaemonUnprocessable,
+            )
+
+            try:
+                result = await asyncio.to_thread(
+                    daemon_request, "POST", path, payload, preserve_conflict=True,
+                    preserve_not_found=True, preserve_unprocessable=True)
+            except (DaemonConflict, DaemonNotFound, DaemonRefused,
+                    DaemonUnprocessable) as refused:
+                # 4.2.3: an unknown, settled or forbidden case is an answer;
+                # retrying cannot help, so the agent is told not to.
+                return {"success": False, "retryable": False,
+                        "error": f"the correction review was refused: {refused}"}
             if isinstance(result, dict) and result.get("success"):
                 return result
+            if isinstance(result, dict):
+                # 4.2.3: the daemon answered and refused (an unknown or settled
+                # case). Retrying cannot help, so say so and pass its reason on.
+                detail = result.get("detail") or result.get("error") or ""
+                if isinstance(detail, dict):
+                    detail = detail.get("message") or detail.get("code") or ""
+                return {
+                    "success": False,
+                    "retryable": False,
+                    "error": "the correction review was refused"
+                             + (f": {detail}" if detail else ""),
+                }
             return {
                 "success": False,
                 "retryable": True,
-                "error": "resident daemon rejected the correction review",
+                "error": "resident daemon did not answer the correction review",
             }
         except Exception:
             logger.exception("review_correction failed")

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/mcp/tools_evolution.py

@@ -265,5 +265,11 @@ def register_evolution_tools(server, get_engine: Callable) -> None:
                 "tree": tree,
             }
         except Exception as exc:
+            # 4.2.3: a store where evolution never ran has no log table yet;
+            # that is an empty lineage, not a failure.
+            if "no such table" in str(exc):
+                return {"success": True, "lineage": [], "lineage_count": 0, "tree": {},
+                        "note": "No skill has evolved in this profile yet."}
             logger.debug("skill_lineage failed: %s", exc)
-            return {"lineage": [], "lineage_count": 0, "tree": {}, "error": str(exc)}
+            return {"success": False, "lineage": [], "lineage_count": 0, "tree": {},
+                    "error": str(exc)}

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/server/recall_core.py

@@ -56,6 +56,12 @@ def recall_budget_s() -> float:
 #: A call budget is never allowed below this: the keyword fallback still has to run.
 RECALL_BUDGET_FLOOR_S = 2.0
 
+#: A short-clock recall still running after this long (or 40% of its budget,
+#: if sooner) starts the keyword fallback beside it. A small store answers
+#: first and never pays for it; on 27k facts the fallback measured 0.07-0.5 s
+#: idle and 1.3 s on a loaded machine, so it needs most of the budget.
+HEDGE_AFTER_S = 1.0
+
 
 def parse_budget_s(raw: Any) -> float | None:
     """A caller's recall budget in seconds, or ``None`` when absent or unusable.
@@ -299,8 +305,14 @@ def is_explicit(call: RecallCall) -> bool:
     ``fast`` (skip the internal agentic round, the agent default) does not
     matter here: an agent's recall tool is fast and explicit.
     """
-    return not (call.skip_answer_check or call.context_load
-                or call.budget_s is not None or call.depth == 0)
+    # depth 0 is how much to show, and a context load (session_init, slm
+    # context) only skips the answer check: someone waits on both, so both
+    # get the full clock. Only the prompt hook (answer_check=skip, it blocks
+    # typing) and a relayed budget keep the short one (4.2.3: on a 27k-fact
+    # store both came back empty, every channel "timeout").
+    if call.context_load:
+        return call.budget_s is None
+    return not (call.skip_answer_check or call.budget_s is not None)
 
 
 def busy(call: RecallCall, snapshot: Any, *, retry_after_s: float) -> dict:
@@ -428,14 +440,29 @@ async def _run_admitted(engine, call, app_state, runtime, snapshot,
                 remaining_deadline(max(0.0, worker_deadline - loop.time()) + EXPLICIT_RUN_GRACE_S),
                 remaining_deadline(EXPLICIT_CEILING_S) - 1.0)
             public_deadline = max(public_deadline, run_deadline + 0.15)
-        response = await wait_until(future, run_deadline)
+        hedge = None
+        if explicit:
+            response = await wait_until(future, run_deadline)
+        else:
+            # Delayed hedge: a short-clock run still going near its deadline
+            # starts the keyword fallback beside it, so an overrun answers with
+            # keyword matches instead of nothing (4.2.3: a 2.5 s recall on a big
+            # store left the fallback ~0.2 s and the prompt hook injected none).
+            left = max(0.0, worker_deadline - loop.time())
+            response = await wait_until(
+                future, loop.time() + min(HEDGE_AFTER_S, left * 0.4))
+            if response is None:
+                hedge = submit(loop, hold, runtime, app_state, engine,
+                               _fallback, engine, call, app_state)
+                response = await wait_until(future, worker_deadline)
         if response is None:
             if explicit:
                 logger.warning("Explicit recall overran its run budget; reporting busy")
                 return busy(call, snapshot, retry_after_s=_retry_after())
-            logger.warning("Recall deadline reached; trying unjudged keyword fallback")
+            logger.warning("Recall deadline reached; serving unjudged keyword fallback")
             return await _bounded_fallback(
-                    engine, call, app_state, public_deadline, snapshot, hold, runtime)
+                    engine, call, app_state, public_deadline, snapshot, hold, runtime,
+                    started=hedge)
         minimal = _minimal_envelope(call, response, snapshot)
         envelope = submit(loop, hold, runtime, app_state, engine, _envelope,
                           engine, call, response, snapshot)
@@ -476,12 +503,14 @@ def _minimal_envelope(call, response, snapshot):
     }
 
 
-async def _bounded_fallback(engine, call, app_state, deadline, snapshot, hold, runtime):
+async def _bounded_fallback(engine, call, app_state, deadline, snapshot, hold, runtime,
+                            started=None):
     from superlocalmemory.server.recall_deadline import submit, timed_out, unjudged, wait_until
     from superlocalmemory.server.recall_ingress import fallback_deadline
 
-    future = submit(asyncio.get_running_loop(), hold, runtime, app_state, engine,
-                    _fallback, engine, call, app_state)
+    future = started if started is not None else submit(
+        asyncio.get_running_loop(), hold, runtime, app_state, engine,
+        _fallback, engine, call, app_state)
     # The fallback keeps its floor even after a budget shortened below it.
     result = await wait_until(future, fallback_deadline(deadline, RECALL_BUDGET_FLOOR_S))
     return unjudged(result) if result is not None else timed_out(call, snapshot)

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/server/recall_fallback.py

@@ -23,7 +23,6 @@ from __future__ import annotations
 
 import logging
 import re
-import time
 from typing import Any
 
 from superlocalmemory.retrieval import channel_status as chstat
@@ -203,7 +202,6 @@ def recall_keyword_fallback(
     tag_scope: dict | None = None
     from superlocalmemory.storage.models import RecallResponse
     response = RecallResponse(source_fidelity_status="degraded")
-    deadline = time.monotonic() + fidelity_budget_s
     try:
         db = engine._db
         pid = profile_id or engine.profile_id
@@ -220,10 +218,15 @@ def recall_keyword_fallback(
             candidates, facet_filter_error, project_scope, tag_scope = _apply_facets(
                 engine, db, pid, candidates, facets)
         from superlocalmemory.retrieval.fallback_fidelity import admit_fallback
-        candidates, response = admit_fallback(
-            candidates, db, pid, budget_s=deadline - time.monotonic())
+        # 4.2.3: choose what would be shown first, then source-check only that,
+        # on the check's own budget. Checking all ~40 over-fetched rows, on a
+        # budget the keyword query had already spent (0.59 s on 27k facts),
+        # timed out and withheld every row: the fallback answered nothing.
         if not has_facets:
             candidates, tier, note = _tiered(query, candidates, show_weak, include_related)
+            candidates = candidates[:limit]
+        candidates, response = admit_fallback(
+            candidates, db, pid, budget_s=fidelity_budget_s)
         for pos, d in enumerate(candidates[:limit], start=1):
             results.append({
                 "fact_id": d.get("fact_id"),

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/server/recall_ingress.py

@@ -15,8 +15,9 @@ def is_recall(scope):
 def explicit_scope(scope):
     """The request is a question someone waits on (``recall_core.is_explicit``).
 
-    Hook and context loads say so in the query string (``answer_check=skip`` or
-    ``context``, ``depth=0``, or a ``budget_s``); they keep the short clock.
+    The prompt hook says so in the query string (``answer_check=skip``), a
+    relayed call with a ``budget_s``; they keep the short clock. ``depth=0``
+    and ``answer_check=context`` (session_init) are waited on too (4.2.3).
     """
     if scope.get("path", "").rstrip("/") != "/recall":
         return True
@@ -24,9 +25,9 @@ def explicit_scope(scope):
 
     params = parse_qs((scope.get("query_string") or b"").decode("latin-1"))
     first = {k: (v[0] if v else "") for k, v in params.items()}
-    if first.get("answer_check", "").strip().lower() in ("skip", "context"):
+    if first.get("answer_check", "").strip().lower() == "skip":
         return False
-    if first.get("budget_s", "").strip() or first.get("depth", "").strip() == "0":
+    if first.get("budget_s", "").strip():
         return False
     return True
 

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/server/remote_profile_binding.py

@@ -126,6 +126,9 @@ ACTIVE_ONLY_TOOLS: dict[str, str] = {}
 NEUTRAL_ARGUMENTS: frozenset[str] = frozenset({
     # ack: message ids a web app received; the broker acknowledges only the caller's own mail.
     "ack",
+    # budget_s (4.2.3): seconds a turn-blocking recall may take; it only shortens
+    # the recall's own clock and selects nothing.
+    "budget_s",
     "about", "action", "agent_id", "as_of", "assertion_id", "case_id", "category",
     "ccr_id", "content", "context", "correction", "depth", "duration_ms", "event_type",
     "event_valid_until", "expected_version", "fact_id", "fact_ids", "fast", "feedback",

{superlocalmemory-4.2.2 → superlocalmemory-4.2.3}/src/superlocalmemory/ui/css/design-system.css

@@ -714,6 +714,19 @@ button { font-family: inherit; }
 .od-whatsnew-tile h4 { font-size: 14.5px; font-weight: 640; margin: 0; }
 .od-whatsnew-tile p { font-size: 13.5px; line-height: 1.5; margin: 0; flex: 1 1 auto; }
 .od-whatsnew-dev { font-size: 12px; margin: 12px 0 0; }
+/* First-run Welcome card (od-welcome.js) */
+#od-welcome:empty { display: none; }
+#od-welcome { margin-bottom: var(--space-3, 16px); }
+.od-welcome > h3 { font-size: 16px; font-weight: 640; margin: 0 0 14px; }
+.od-welcome-row { display: flex; flex-direction: column; align-items: flex-start; gap: 8px; padding: 14px 16px; margin-bottom: 12px; background: var(--card-2); border: 1px solid var(--border); border-radius: var(--r-md); }
+.od-welcome-row-head { display: flex; align-items: center; gap: 8px; flex-wrap: wrap; }
+.od-welcome-row h4 { font-size: 14.5px; font-weight: 640; margin: 0; }
+.od-welcome-row p { font-size: 13.5px; line-height: 1.5; margin: 0; }
+.od-welcome-list { margin: 0; padding-left: 20px; font-size: 13.5px; line-height: 1.6; }
+.od-welcome-prompt { width: 100%; margin: 0; padding: 10px 12px; font-family: var(--font-mono, ui-monospace, monospace); font-size: 12.5px; line-height: 1.5; white-space: pre-wrap; overflow-wrap: anywhere; color: var(--fg); background: var(--card); border: 1px solid var(--border); border-radius: var(--r-md); user-select: all; }
+.od-welcome-actions { display: flex; align-items: center; gap: 10px; flex-wrap: wrap; }
+.od-welcome-note:empty { display: none; }
+.od-welcome-foot { display: flex; justify-content: flex-end; }
 .od-dropzone { display: flex; flex-direction: column; align-items: center; gap: 8px; padding: 28px 16px; text-align: center; border: 2px dashed var(--border); border-radius: var(--r-md); background: var(--card-2); }
 .od-dropzone.is-over { border-color: var(--violet); background: var(--violet-soft); }
 .od-dropzone p { margin: 0; font-size: 14px; }