unique-user-memory 2026.32.0.dev3__tar.gz → 2026.32.0.dev4__tar.gz

{unique_user_memory-2026.32.0.dev3 → unique_user_memory-2026.32.0.dev4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: unique-user-memory
-Version: 2026.32.0.dev3
+Version: 2026.32.0.dev4
 Summary: 
 Author: Fabian Schläpfer
 Author-email: Fabian Schläpfer <fabian@unique.ch>
@@ -24,6 +24,7 @@ The package provides:
 
 - `UserMemoryConfig` - Pydantic configuration for the consolidation model, profile token budget, and memory folder.
 - `load_user_memory(...)` - resolves the user's private memory folder, downloads `memory.md`, and enforces the configured token budget. The `language_model` argument is used to tokenize `memory.md` when capping it, so it must be the same effective model the postprocessor uses for consolidation (see Integration below). Returns a `UserMemoryState` with the profile text and scope id.
+- `profile_body(...)` - strips the YAML frontmatter and returns only the Markdown body. Use it whenever the profile is shown to a model; the frontmatter is bookkeeping for consolidation.
 - `UserMemoryMessageLogger` - emits chat Steps (MessageLogs) for load and update, including typed `UserMemory` detail entries the chat frontend renders as a badge that opens Settings → Context Memory. Frontends that do not know the entry type render nothing, so the entries are safe to emit in any deploy order.
 - `UserMemoryPostprocessor` - runs after the assistant response, consolidates the latest turn into the profile, and uploads the updated `memory.md`.
 
@@ -34,7 +35,7 @@ The memory file is intentionally small and structured. It is rewritten as a full
 1. The orchestrator enables memory when `space.allow_user_memory` is true.
 2. The orchestrator emits a **Loading context memory** Step, then `load_user_memory(...)` resolves the pre-provisioned root folder, ensures a private child folder for the current user, and downloads `/user-memory/<user_id>/memory.md` if it exists.
 3. When load returns a `UserMemoryState`, that Step is completed with a **Context memory** detail entry (`type: UserMemory`) that the chat frontend renders as a badge opening Settings → Context Memory. A successful `None` return (soft skip) completes the Step without the entry; a raised exception marks the Step failed.
-4. If memory was loaded, its text is passed into the agent context for the current turn.
+4. If memory was loaded, `profile_body(...)` of its text is passed into the agent context for the current turn — the prompt only gets the Markdown body, while the postprocessor keeps the full file because it needs the frontmatter to carry `turn_count` forward.
 5. `UserMemoryPostprocessor` runs after the assistant response.
 6. The package asks the configured language model to either return `NOOP` or a complete rewritten profile.
 7. If a rewrite runs, an **Updating your memory** Step is shown while consolidating (no settings entry yet).
@@ -113,7 +114,7 @@ Typical orchestration code loads memory before the agent loop and registers the
 
 ```python
 from unique_toolkit.agentic.message_log_manager.service import MessageStepLogger
-from unique_user_memory.user_memory import load_user_memory
+from unique_user_memory.user_memory import load_user_memory, profile_body
 from unique_user_memory.user_memory_message_log import UserMemoryMessageLogger
 from unique_user_memory.user_memory_postprocessor import UserMemoryPostprocessor
 
@@ -157,7 +158,9 @@ finally:
 
 if load_succeeded and user_memory_state is not None:
     await memory_message_step_logger.log_loading_complete(with_settings_entry=True)
-    user_memory_text = user_memory_state.text
+    # The postprocessor keeps the full file (it needs the frontmatter to
+    # carry turn_count forward); the prompt only gets the Markdown body.
+    user_memory_text = profile_body(user_memory_state.text)
     postprocessor_manager.add_postprocessor(
         UserMemoryPostprocessor(
             config=user_memory_config,

{unique_user_memory-2026.32.0.dev3 → unique_user_memory-2026.32.0.dev4}/README.md

@@ -10,6 +10,7 @@ The package provides:
 
 - `UserMemoryConfig` - Pydantic configuration for the consolidation model, profile token budget, and memory folder.
 - `load_user_memory(...)` - resolves the user's private memory folder, downloads `memory.md`, and enforces the configured token budget. The `language_model` argument is used to tokenize `memory.md` when capping it, so it must be the same effective model the postprocessor uses for consolidation (see Integration below). Returns a `UserMemoryState` with the profile text and scope id.
+- `profile_body(...)` - strips the YAML frontmatter and returns only the Markdown body. Use it whenever the profile is shown to a model; the frontmatter is bookkeeping for consolidation.
 - `UserMemoryMessageLogger` - emits chat Steps (MessageLogs) for load and update, including typed `UserMemory` detail entries the chat frontend renders as a badge that opens Settings → Context Memory. Frontends that do not know the entry type render nothing, so the entries are safe to emit in any deploy order.
 - `UserMemoryPostprocessor` - runs after the assistant response, consolidates the latest turn into the profile, and uploads the updated `memory.md`.
 
@@ -20,7 +21,7 @@ The memory file is intentionally small and structured. It is rewritten as a full
 1. The orchestrator enables memory when `space.allow_user_memory` is true.
 2. The orchestrator emits a **Loading context memory** Step, then `load_user_memory(...)` resolves the pre-provisioned root folder, ensures a private child folder for the current user, and downloads `/user-memory/<user_id>/memory.md` if it exists.
 3. When load returns a `UserMemoryState`, that Step is completed with a **Context memory** detail entry (`type: UserMemory`) that the chat frontend renders as a badge opening Settings → Context Memory. A successful `None` return (soft skip) completes the Step without the entry; a raised exception marks the Step failed.
-4. If memory was loaded, its text is passed into the agent context for the current turn.
+4. If memory was loaded, `profile_body(...)` of its text is passed into the agent context for the current turn — the prompt only gets the Markdown body, while the postprocessor keeps the full file because it needs the frontmatter to carry `turn_count` forward.
 5. `UserMemoryPostprocessor` runs after the assistant response.
 6. The package asks the configured language model to either return `NOOP` or a complete rewritten profile.
 7. If a rewrite runs, an **Updating your memory** Step is shown while consolidating (no settings entry yet).
@@ -99,7 +100,7 @@ Typical orchestration code loads memory before the agent loop and registers the
 
 ```python
 from unique_toolkit.agentic.message_log_manager.service import MessageStepLogger
-from unique_user_memory.user_memory import load_user_memory
+from unique_user_memory.user_memory import load_user_memory, profile_body
 from unique_user_memory.user_memory_message_log import UserMemoryMessageLogger
 from unique_user_memory.user_memory_postprocessor import UserMemoryPostprocessor
 
@@ -143,7 +144,9 @@ finally:
 
 if load_succeeded and user_memory_state is not None:
     await memory_message_step_logger.log_loading_complete(with_settings_entry=True)
-    user_memory_text = user_memory_state.text
+    # The postprocessor keeps the full file (it needs the frontmatter to
+    # carry turn_count forward); the prompt only gets the Markdown body.
+    user_memory_text = profile_body(user_memory_state.text)
     postprocessor_manager.add_postprocessor(
         UserMemoryPostprocessor(
             config=user_memory_config,

{unique_user_memory-2026.32.0.dev3 → unique_user_memory-2026.32.0.dev4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "unique_user_memory"
-version = "2026.32.0.dev3"
+version = "2026.32.0.dev4"
 description = ""
 authors = [
     { name = "Fabian Schläpfer", email = "fabian@unique.ch" },

{unique_user_memory-2026.32.0.dev3 → unique_user_memory-2026.32.0.dev4}/unique_user_memory/tests/test_user_memory.py

@@ -21,6 +21,7 @@ from unique_user_memory.user_memory import (
     enforce_token_cap,
     ensure_user_memory_folder,
     fit_user_memory,
+    profile_body,
     should_consolidate_memory,
     upload_user_memory,
 )
@@ -46,6 +47,37 @@ def test_memory_profile_keeps_follow_up_tasks_but_excludes_open_questions() -> N
     assert "Concrete future tasks the user intends to complete" in gate_prompt
 
 
+def test_profile_body_strips_frontmatter() -> None:
+    # The orchestrator renders this into the system prompt, where the
+    # bookkeeping fields are noise for the model.
+    content = (
+        "---\n"
+        "user_id: 233737684428787846\n"
+        "strategy: codex\n"
+        "schema_version: 1\n"
+        "last_updated: 2026-07-27T10:20:00+00:00\n"
+        "turn_count: 83\n"
+        "---\n\n"
+        "# User Memory\n\n## Identity\n- Andreas\n"
+    )
+
+    body = profile_body(content)
+
+    assert body == "# User Memory\n\n## Identity\n- Andreas"
+    for field in ("user_id:", "strategy:", "schema_version:", "turn_count:"):
+        assert field not in body
+
+
+def test_profile_body_leaves_frontmatterless_profile_untouched() -> None:
+    content = "# User Memory\n\n## Identity\n- Andreas"
+
+    assert profile_body(content) == content
+
+
+def test_profile_body_returns_empty_for_frontmatter_only_profile() -> None:
+    assert profile_body("---\nuser_id: 42\nturn_count: 0\n---\n") == ""
+
+
 def test_enforce_token_cap_truncates_long_content() -> None:
     content = "\n\n".join(f"paragraph {index} " + "word " * 40 for index in range(50))
 

{unique_user_memory-2026.32.0.dev3 → unique_user_memory-2026.32.0.dev4}/unique_user_memory/user_memory.py

@@ -63,7 +63,14 @@ _FRONTMATTER_RE = re.compile(r"^---\n.*?\n---\n", re.DOTALL)
 _TURN_COUNT_RE = re.compile(r"^turn_count:\s*(\d+)\s*$", re.MULTILINE)
 
 
-def _profile_body(content: str) -> str:
+def profile_body(content: str) -> str:
+    """Return the profile without its YAML frontmatter.
+
+    The frontmatter is bookkeeping for the consolidation pass (turn count,
+    schema version, timestamps); consumers that show the profile to a model --
+    consolidation prompts and the orchestrator system prompt -- only want the
+    Markdown body.
+    """
     return _FRONTMATTER_RE.sub("", content, count=1).strip()
 
 
@@ -195,7 +202,7 @@ async def condense_user_memory(
     condensed profile, or ``None`` when the call fails or the output does
     not look like a profile (the caller then falls back to a hard cut).
     """
-    body = _profile_body(content)
+    body = profile_body(content)
     current_tokens = count_tokens(content=body, language_model=language_model)
     target_tokens = max(1, int(max_tokens * _CONDENSE_TARGET_RATIO))
 
@@ -265,7 +272,7 @@ async def condense_user_memory(
         )
         return None
 
-    candidate = _profile_body(_strip_code_fences(raw))
+    candidate = profile_body(_strip_code_fences(raw))
     if not _is_well_formed_profile(candidate):
         logger.warning(
             "[user-memory] condense output did not look like a profile (%d chars)",
@@ -826,7 +833,7 @@ async def _rewrite_user_memory(
             ),
             LanguageModelUserMessage(
                 content=consolidation_user_prompt(
-                    existing_memory=_profile_body(safe_current),
+                    existing_memory=profile_body(safe_current),
                     user_message=_sanitize_for_xml_context(user_message or ""),
                     assistant_message=_sanitize_for_xml_context(
                         assistant_message or ""
@@ -883,7 +890,7 @@ async def _rewrite_user_memory(
         logger.info("[user-memory] consolidation NOOP - keeping existing memory")
         return safe_current
 
-    candidate_body = _profile_body(_strip_code_fences(raw))
+    candidate_body = profile_body(_strip_code_fences(raw))
     if not _is_well_formed_profile(candidate_body):
         logger.warning(
             "[user-memory] LLM output did not look like a profile (%d chars)",
@@ -891,7 +898,7 @@ async def _rewrite_user_memory(
         )
         return safe_current
 
-    if safe_current and candidate_body == _profile_body(safe_current):
+    if safe_current and candidate_body == profile_body(safe_current):
         logger.debug("[user-memory] memory body unchanged - skipping update")
         return safe_current