dev-memory-cli 0.22.1 → 0.23.0

package/scripts/hooks/_common.py

@@ -41,6 +41,7 @@ CONTEXT_SCRIPT = PACKAGE_ROOT / "lib" / "dev_memory_context.py"
 CAPTURE_SCRIPT = PACKAGE_ROOT / "lib" / "dev_memory_capture.py"
 SUMMARY_WORKER_SCRIPT = PACKAGE_ROOT / "scripts" / "hooks" / "session_summary_worker.py"
 DEFAULT_CONFIG_PATH = Path(os.environ.get("DEV_MEMORY_CONFIG_PATH", "~/.dev-memory/config.json")).expanduser()
+DEV_MEMORY_HOME = Path(os.environ.get("DEV_MEMORY_HOME", "~/.dev-memory")).expanduser()
 WORKSPACE_CONFIG_NAMES = (".dev-memory-workspace.json", ".dev-assets-workspace.json")
 
 
@@ -121,6 +122,28 @@ def read_hook_input():
         return {"raw": raw[:4000]}
 
 
+def register_session_scan_candidate(hook_input):
+    """Persist a lightweight Codex Stop hint without starting summarization."""
+    if not isinstance(hook_input, dict):
+        return None
+    session_id = hook_session_id(hook_input)
+    transcript_path = _hook_transcript_path(hook_input)
+    if not session_id and not transcript_path:
+        return None
+    source = session_id or transcript_path
+    digest = hashlib.sha1(str(source).encode("utf-8")).hexdigest()[:20]
+    path = DEV_MEMORY_HOME / "jobs" / "session-scan" / "candidates" / f"{digest}.json"
+    _atomic_write_json(path, {
+        "schema_version": 1,
+        "event": "Stop",
+        "registered_at": now_iso(),
+        "session_id": session_id,
+        "transcript_path": transcript_path,
+        "cwd": _first_string(hook_input.get("cwd"), str(REPO_ROOT)),
+    })
+    return path
+
+
 def resolve_assets_for(repo_root):
     """Resolve asset paths for an explicit repo root (workspace-mode friendly)."""
     repo_root_str = str(repo_root)
@@ -219,7 +242,7 @@ def extract_section(path, title):
     return None if is_placeholder(body) else body
 
 
-def extract_repo_file_body(path):
+def extract_repo_file_body(path, *, newest_first=False):
     """Extract full body of a repo-level file, skipping the H1 title and
     ``## 仓库`` metadata section. Placeholder-only sections within the file
     are dropped individually so that real content in sibling sections is
@@ -235,6 +258,11 @@ def extract_repo_file_body(path):
     for sec in sections:
         sec_body = strip_managed_markers(sec).strip()
         if sec_body and not is_placeholder(sec_body):
+            if newest_first and sec_body.startswith("## "):
+                heading, separator, body = sec_body.partition("\n")
+                if separator and body.strip():
+                    blocks = _split_recent_blocks(body)
+                    sec_body = heading + "\n\n" + "\n\n".join(blocks)
             kept.append(sec_body)
     body = "\n\n".join(kept).strip()
     return body or None
@@ -443,11 +471,14 @@ def _brief_section_keys(profile):
     return tuple(_BRIEF_PROFILES.get(profile, _BRIEF_PROFILES["standard"]).keys())
 
 
-def _extract_sections(paths, keys):
+def _extract_sections(paths, keys, *, repo_newest_first=False):
     out = []
     for file_key, title in keys:
         if title is None:
-            body = extract_repo_file_body(paths[file_key])
+            body = extract_repo_file_body(
+                paths[file_key],
+                newest_first=repo_newest_first and file_key in {"repo_decisions", "repo_glossary"},
+            )
             display_title = _REPO_DISPLAY_TITLES.get(file_key, paths[file_key].stem)
         else:
             body = extract_section(paths[file_key], title)
@@ -459,19 +490,18 @@ def _extract_sections(paths, keys):
 
 def _build_context_from_assets(assets, *, full=True, heading=None, brief_profile="standard"):
     if not assets["branch_dir"].exists():
-        # v2: capture lazy-inits on first write, so branch_dir typically
-        # exists after any real interaction. Missing here just means no
-        # write has happened yet — no need to push setup.
+        # Memory is opt-in. A missing branch directory stays silent in
+        # workspace briefs; the primary repo gets a concise init hint.
         if heading is None:
             return (
                 "当前仓库+分支还没有 dev-memory 记忆。"
-                "下一次 `dev-memory-capture` 写入时会自动 lazy init；现有结论若值得记一笔，直接走 capture。"
+                "需要启用时运行 `dev-memory-cli init`；之后由定期 session-scan 沉淀会话语义。"
             )
         return None
 
     paths = assets["paths"]
     keys = _FULL_SECTION_KEYS if full else _brief_section_keys(brief_profile)
-    sections = _extract_sections(paths, keys)
+    sections = _extract_sections(paths, keys, repo_newest_first=full)
     max_lines, max_chars = (8, 700) if full else (3, 200)
     brief_limits = _BRIEF_PROFILES.get(brief_profile, _BRIEF_PROFILES["standard"])
 
@@ -586,7 +616,7 @@ def _build_context_from_assets(assets, *, full=True, heading=None, brief_profile
                 "",
                 f"归档分支查询：`grep -r 'KEYWORD' {archive_dir}/` （体量大时派 Task 子 agent）",
             ])
-        footer_lines.extend(["", "新决策 / 进展 / 阻塞 → `dev-memory-capture` 写入。"])
+        footer_lines.extend(["", "新增会话语义由定期 session-scan 沉淀；主动查旧记忆使用 `dev-memory-read`。"])
         parts.append("\n".join(footer_lines))
 
     return "\n\n".join(parts)
@@ -898,9 +928,14 @@ def build_summary_input(job_path):
     job = json.loads(Path(job_path).read_text(encoding="utf-8"))
     return extract_core_payload(
         job,
-        max_messages=_int_env("DEV_MEMORY_SESSION_SUMMARY_MAX_MESSAGES", 30),
-        max_message_chars=_int_env("DEV_MEMORY_SESSION_SUMMARY_MAX_MESSAGE_CHARS", 1600),
+        max_messages=_int_env("DEV_MEMORY_SESSION_SUMMARY_MAX_MESSAGES", 0),
+        max_message_chars=_int_env("DEV_MEMORY_SESSION_SUMMARY_MAX_MESSAGE_CHARS", 0),
         max_memory_chars=_int_env("DEV_MEMORY_SESSION_SUMMARY_MAX_MEMORY_CHARS", 4000),
+        since_size=(
+            ((job.get("previous_job") or {}).get("processed") or {}).get("transcript_size", 0)
+            if isinstance(job.get("previous_job"), dict)
+            else 0
+        ),
     )
 
 
@@ -918,7 +953,8 @@ def build_summary_prompt(job_path, summary_input=None, summary_input_path=None):
         summary_input = build_summary_input(job_path)
     summary_input_json = json.dumps(summary_input, ensure_ascii=False, indent=2)
     input_path_line = f"- summary input JSON: {summary_input_path}\n" if summary_input_path else ""
-    return f"""你是 dev-memory 的后台会话总结 worker。
+    return f"""DEV_MEMORY_INTERNAL_SESSION_SUMMARY_V1
+你是 dev-memory 的后台会话总结 worker。
 
 输入：
 - job JSON: {job_path}

package/scripts/hooks/stop.py

@@ -5,13 +5,19 @@ from _common import (
     is_workspace_mode,
     log,
     maybe_record_head,
+    read_hook_input,
     record_head_all_repos,
+    register_session_scan_candidate,
     resolve_assets,
 )
 
 
 def main():
+    hook_input = read_hook_input()
     try:
+        candidate = register_session_scan_candidate(hook_input)
+        if candidate:
+            log(f"[dev-memory][Stop] registered session scan candidate {candidate}")
         if is_no_git_mode():
             log("[dev-memory][Stop] no-git mode: nothing to record (no HEAD)")
             return 0

package/suite-manifest.json

@@ -1,13 +1,13 @@
 {
   "suite_name": "dev-memory-skill-suite",
   "skills": [
-    "dev-memory-read",
+    "dev-memory-read"
+  ],
+  "legacy_skills": [
     "dev-memory-setup",
     "dev-memory-capture",
     "dev-memory-graduate",
-    "dev-memory-tidy"
-  ],
-  "legacy_skills": [
+    "dev-memory-tidy",
     "branch-context",
     "branch-context-setup",
     "dev-asset-setup",