code-data-ark 2.0.2__tar.gz → 2.0.3__tar.gz

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: code-data-ark
-Version: 2.0.2
+Version: 2.0.3
 Summary: Code Data Ark — local observability and intelligence platform for VS Code + Copilot Chat sessions
 Project-URL: Homepage, https://github.com/goCosmix/cda
 Project-URL: Repository, https://github.com/goCosmix/cda.git
@@ -121,39 +121,41 @@ make install-dev
 
 ## ⚡ Quick Start
 
-1. **Initialize the database**
+1. **Install**
 
 ```bash
-cda sync
+pip install code-data-ark
 ```
 
-2. **Start the watcher daemon**
+2. **Initialize — create `~/.cda/` and validate your VS Code data path**
 
 ```bash
-cda watch start
+cda init
 ```
 
-3. **Inspect the PMF runtime services**
+3. **Ingest all VS Code session data**
 
 ```bash
-cda pmf services
+cda sync
 ```
 
-4. **Build semantic intelligence**
+4. **Start the live watcher daemon**
 
 ```bash
-cda embed build
+cda watch start
 ```
 
-4. **Start the web UI**
+5. **Open the web dashboard**
 
 ```bash
-cda ui start
+cda serve   # → http://127.0.0.1:10001
 ```
 
-5. **Open your browser**
+6. **Build semantic intelligence** (optional, requires `sentence-transformers`)
 
-Visit `http://127.0.0.1:10001`
+```bash
+cda embed build
+```
 
 ## 🌐 Web UI
 

code_data_ark-2.0.3/cda/kernel/paths.py

@@ -0,0 +1,54 @@
+"""
+cda.kernel.paths — canonical path resolution for Code Data Ark.
+
+CDA_HOME is the single root for all runtime state (DB, PID files, logs,
+queue, PMF runtime).  It is resolved exactly once at import time via:
+
+  1. CDA_HOME environment variable (absolute path)
+  2. ~/.cda/  (default — survives pip install, editable install, CI)
+
+Pipeline stages and the CLI all import from here so every module agrees
+on the same paths regardless of where the package is installed.
+"""
+
+import os
+from pathlib import Path
+
+# ── home resolution ──────────────────────────────────────────────────────────
+
+
+def get_cda_home() -> Path:
+    """Return the CDA home directory, creating it if it doesn't exist."""
+    env = os.environ.get("CDA_HOME")
+    if env:
+        home = Path(env).expanduser().resolve()
+    else:
+        home = Path.home() / ".cda"
+    home.mkdir(parents=True, exist_ok=True)
+    return home
+
+
+# ── canonical paths (module-level constants, computed once) ─────────────────
+
+CDA_HOME   = get_cda_home()
+LOCAL_DIR  = CDA_HOME                         # CDA_HOME *is* the local root
+DATA_DIR   = CDA_HOME / "data"
+RUN_DIR    = CDA_HOME / "run"
+LOG_DIR    = CDA_HOME / "logs"
+QUEUE_DIR  = CDA_HOME / "queue"
+PMF_DIR    = CDA_HOME / "pmf"
+CONFIG_DIR = CDA_HOME / "config"
+
+DB_PATH        = DATA_DIR / "cda.db"
+PID_FILE       = RUN_DIR / "watcher.pid"
+UI_PID_FILE    = RUN_DIR / "ui.pid"
+UI_LOG_FILE    = LOG_DIR / "ui.log"
+POLICY_FILE    = CONFIG_DIR / "policy.txt"
+PMF_LOG_DIR    = PMF_DIR / "logs"
+RUNTIME_FILE   = PMF_DIR / "runtime.json"
+
+
+def ensure_dirs() -> None:
+    """Create all runtime directories. Safe to call multiple times."""
+    for d in (DATA_DIR, RUN_DIR, LOG_DIR, QUEUE_DIR, PMF_DIR, PMF_LOG_DIR, CONFIG_DIR):
+        d.mkdir(parents=True, exist_ok=True)

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/kernel/pmf_kernel.py

@@ -8,19 +8,16 @@ from dataclasses import dataclass
 from pathlib import Path
 from typing import Dict, List, Optional
 
-ROOT_DIR = Path(__file__).resolve().parent.parent.parent.parent
-LOCAL_DIR = ROOT_DIR / "local"
-PACKAGE_DIR = Path(__file__).resolve().parent
-RUNTIME_FILE = LOCAL_DIR / "pmf" / "runtime.json"
-LOG_DIR = LOCAL_DIR / "pmf" / "logs"
-WATCHER_PID_FILE = LOCAL_DIR / "run" / "watcher.pid"
-UI_PID_FILE = LOCAL_DIR / "run" / "ui.pid"
+from cda.kernel.paths import (
+    LOG_DIR, RUNTIME_FILE, PMF_LOG_DIR,
+    PID_FILE as WATCHER_PID_FILE, UI_PID_FILE, CDA_HOME,
+    ensure_dirs,
+)
+
 DEFAULT_HOST = "127.0.0.1"
 DEFAULT_PORT = 10001
 
-(LOCAL_DIR / "data").mkdir(parents=True, exist_ok=True)
-(LOCAL_DIR / "run").mkdir(parents=True, exist_ok=True)
-LOG_DIR.mkdir(parents=True, exist_ok=True)
+ensure_dirs()
 
 
 def now_ts():
@@ -58,7 +55,7 @@ class ServiceSpec:
             ]
 
         if self.service_id == "watcher":
-            return [sys.executable, str(PACKAGE_DIR.parent / "pipeline" / "watcher.py")]
+            return [sys.executable, "-m", "cda.pipeline.watcher"]
 
         if self.command is not None:
             return list(self.command)
@@ -72,9 +69,9 @@ SERVICE_SPECS: Dict[str, ServiceSpec] = {
         label="Watcher Daemon",
         service_type="daemon",
         description="Live VS Code data watcher and incremental ingest process.",
-        cwd=ROOT_DIR,
+        cwd=CDA_HOME,
         pid_file=WATCHER_PID_FILE,
-        log_file=LOCAL_DIR / "logs" / "watcher.log",
+        log_file=LOG_DIR / "watcher.log",
         allowed_actions=["start", "stop", "restart", "status"],
     ),
     "ui": ServiceSpec(
@@ -82,9 +79,9 @@ SERVICE_SPECS: Dict[str, ServiceSpec] = {
         label="Web UI",
         service_type="daemon",
         description="Local web dashboard for Ark runtime and session analytics.",
-        cwd=ROOT_DIR,
+        cwd=CDA_HOME,
         pid_file=UI_PID_FILE,
-        log_file=LOCAL_DIR / "logs" / "ui.log",
+        log_file=LOG_DIR / "ui.log",
         allowed_actions=["start", "stop", "restart", "status"],
     ),
     "sync": ServiceSpec(
@@ -92,9 +89,9 @@ SERVICE_SPECS: Dict[str, ServiceSpec] = {
         label="Full Sync",
         service_type="task",
         description="Full ingest and rebuild pipeline for Ark data.",
-        command=[sys.executable, str(PACKAGE_DIR.parent / "pipeline" / "ingest.py")],
-        cwd=ROOT_DIR,
-        log_file=LOG_DIR / "sync.log",
+        command=[sys.executable, "-m", "cda.pipeline.ingest"],
+        cwd=CDA_HOME,
+        log_file=PMF_LOG_DIR / "sync.log",
         allowed_actions=["start", "status"],
     ),
     "reconstruct": ServiceSpec(
@@ -102,9 +99,9 @@ SERVICE_SPECS: Dict[str, ServiceSpec] = {
         label="Reconstruct",
         service_type="task",
         description="Reconstruct conversations and rebuild the full text search index.",
-        command=[sys.executable, str(PACKAGE_DIR.parent / "pipeline" / "reconstruct.py")],
-        cwd=ROOT_DIR,
-        log_file=LOG_DIR / "reconstruct.log",
+        command=[sys.executable, "-m", "cda.pipeline.reconstruct"],
+        cwd=CDA_HOME,
+        log_file=PMF_LOG_DIR / "reconstruct.log",
         allowed_actions=["start", "status"],
     ),
     "embed-build": ServiceSpec(
@@ -112,9 +109,9 @@ SERVICE_SPECS: Dict[str, ServiceSpec] = {
         label="Embed Build",
         service_type="task",
         description="Build semantic embeddings and session intelligence.",
-        command=[sys.executable, str(PACKAGE_DIR.parent / "pipeline" / "embed.py"), "build"],
-        cwd=ROOT_DIR,
-        log_file=LOG_DIR / "embed.log",
+        command=[sys.executable, "-m", "cda.pipeline.embed", "build"],
+        cwd=CDA_HOME,
+        log_file=PMF_LOG_DIR / "embed.log",
         allowed_actions=["start", "status"],
     ),
 }
@@ -263,8 +260,8 @@ class PMFKernel:
         with open(log_file, "a") as fh:
             proc = subprocess.Popen(
                 command,
-                cwd=spec.cwd or ROOT_DIR,
-                env={**os.environ, **(spec.env or {}), "PYTHONPATH": str(ROOT_DIR)},
+                cwd=spec.cwd or CDA_HOME,
+                env={**os.environ, **(spec.env or {})},
                 stdout=fh,
                 stderr=fh,
                 preexec_fn=os.setsid if spec.service_type == "daemon" else None,

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/kernel/selfcheck.py

@@ -26,15 +26,12 @@ import subprocess
 import sys
 from pathlib import Path
 
-# ── paths the system knows about itself ─────────────────────────────────────
+from cda.kernel.paths import DB_PATH, PID_FILE, QUEUE_DIR
+
 PACKAGE_DIR  = Path(__file__).resolve().parent
 SOURCE_DIR   = PACKAGE_DIR.parent.parent          # source/  — tracked repo root
-PROJECT_DIR  = PACKAGE_DIR.parent.parent.parent   # repo root — where layers live
-LOCAL_DIR    = PROJECT_DIR / "local"
-DB_PATH      = LOCAL_DIR / "data" / "cda.db"
-PID_FILE     = LOCAL_DIR / "run" / "watcher.pid"
-QUEUE_DIR    = LOCAL_DIR / "queue"
 VERSION_FILE = SOURCE_DIR / "version"
+GIT_ROOT     = SOURCE_DIR.parent                   # repo root — used for git check-ignore
 
 REQUIRED_TABLES = [
     "sessions", "exchanges", "tool_calls", "vfs", "workspaces",
@@ -231,7 +228,7 @@ def check_data_gitignored():
     try:
         result = subprocess.run(
             ["git", "check-ignore", "-q", "local"],
-            cwd=PROJECT_DIR,
+            cwd=GIT_ROOT,
             capture_output=True,
         )
         if result.returncode == 0:

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/pipeline/embed.py

@@ -10,12 +10,10 @@ This stage builds semantic embeddings and mini-intelligence artifacts:
 
 import json
 import sqlite3
-from pathlib import Path
 from typing import Dict, List, Optional, Tuple
 
-ROOT_DIR = Path(__file__).resolve().parent.parent.parent.parent
-LOCAL_DIR = ROOT_DIR / "local"
-DB_PATH = LOCAL_DIR / "data" / "cda.db"
+from cda.kernel.paths import DB_PATH
+
 MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2"
 MAX_EMBED_TEXT = 1400
 

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/pipeline/extract.py

@@ -26,9 +26,7 @@ from datetime import datetime
 from typing import Dict, List, Tuple, DefaultDict
 from collections import defaultdict
 
-ROOT_DIR = Path(__file__).resolve().parent.parent.parent.parent
-LOCAL_DIR = ROOT_DIR / "local"
-DB_PATH = LOCAL_DIR / "data" / "cda.db"
+from cda.kernel.paths import DB_PATH
 
 # ─────────────────────────────────────────────────────────
 # Signal patterns

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/pipeline/ingest.py

@@ -26,6 +26,11 @@ import time
 import logging
 from pathlib import Path
 
+from cda.kernel.paths import DB_PATH, ensure_dirs
+
+# Ensure local dirs are present before writing
+ensure_dirs()
+
 # Set up logging
 logging.basicConfig(
     level=logging.INFO,
@@ -39,9 +44,6 @@ HOME        = Path.home()
 VSCODE_DATA_DIR = Path(os.environ.get("VSCODE_DATA_DIR", HOME / "Library/Application Support/Code/User"))
 VS_STORAGE  = VSCODE_DATA_DIR / "workspaceStorage"
 GLOBAL_MEM  = VSCODE_DATA_DIR / "globalStorage/github.copilot-chat/memory-tool/memories"
-ROOT_DIR = Path(__file__).resolve().parent.parent.parent.parent
-LOCAL_DIR = ROOT_DIR / "local"
-DB_PATH  = LOCAL_DIR / "data" / "cda.db"
 
 # Large index DBs — too big to blob, record path only
 SKIP_BLOB_PATTERNS = ["workspace-chunks.db", "local-index"]

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/pipeline/parse_edits.py

@@ -33,11 +33,7 @@ Edit rounds: len(checkpoints) - 1  (first is always "Initial State")
 import sqlite3
 import gzip
 import json
-from pathlib import Path
-
-ROOT_DIR = Path(__file__).resolve().parent.parent.parent.parent
-LOCAL_DIR = ROOT_DIR / "local"
-DB_PATH = LOCAL_DIR / "data" / "cda.db"
+from cda.kernel.paths import DB_PATH
 
 SCHEMA = """
 CREATE TABLE IF NOT EXISTS edit_sessions (

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/pipeline/reconstruct.py

@@ -19,9 +19,8 @@ import time
 from typing import Optional
 from pathlib import Path
 
-ROOT_DIR = Path(__file__).resolve().parent.parent.parent.parent
-LOCAL_DIR = ROOT_DIR / "local"
-DB_PATH = LOCAL_DIR / "data" / "cda.db"
+from cda.kernel.paths import DB_PATH
+
 NOW_MS  = int(time.time() * 1000)
 
 EXCHANGES_SCHEMA = """

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/pipeline/watcher.py

@@ -38,17 +38,14 @@ except ImportError:
     print("ERROR: watchfiles not installed. Run: pip install watchfiles")
     sys.exit(1)
 
-ROOT_DIR  = Path(__file__).resolve().parent.parent.parent.parent
-LOCAL_DIR = ROOT_DIR / "local"
-DB_PATH   = LOCAL_DIR / "data" / "cda.db"
-PID_FILE  = LOCAL_DIR / "run" / "watcher.pid"
-QUEUE_DIR = LOCAL_DIR / "queue"
+from cda.kernel.paths import DB_PATH, PID_FILE, QUEUE_DIR, LOG_DIR, ensure_dirs
 # Allow override via env var for portability
 VSCODE_DATA_DIR = Path(os.environ.get("VSCODE_DATA_DIR", Path.home() / "Library/Application Support/Code/User"))
 VS_ROOT   = VSCODE_DATA_DIR / "workspaceStorage"
 GLOBAL_MEM = VSCODE_DATA_DIR / "globalStorage/github.copilot-chat/memory-tool/memories"
 
-log_file = LOCAL_DIR / "logs" / "watcher.log"
+ensure_dirs()
+log_file = LOG_DIR / "watcher.log"
 logging.basicConfig(
     level=logging.INFO,
     format="%(asctime)s  %(levelname)-7s  %(message)s",

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/ui/cli.py

@@ -25,6 +25,7 @@ Commands:
   cda pmf restart <service>  Restart a service
   cda pmf logs <service>     Tail service logs
   cda check                  Run a full self-diagnostic. The system checks itself.
+  cda init                   First-run setup — create ~/.cda/ and validate environment
   cda serve                  Start the local web UI on port 10001
   cda sync                   Full re-ingest from disk (rebuilds entire DB)
   cda reconstruct            Re-run reconstruction and FTS rebuild only
@@ -62,22 +63,15 @@ import datetime
 from pathlib import Path
 from cda.pipeline.reconstruct import decompress_vfs
 from cda.kernel.pmf_kernel import PMFKernel, PMFKernelError
+from cda.kernel.paths import (
+    DB_PATH, PID_FILE, UI_PID_FILE, UI_LOG_FILE,
+    QUEUE_DIR, POLICY_FILE, ensure_dirs,
+)
 
 import click
 
-# Package-relative paths
-PACKAGE_DIR = Path(__file__).resolve().parent
-ARK_DIR = PACKAGE_DIR.parent.parent.parent
-LOCAL_DIR = ARK_DIR / "local"
-DB_PATH = LOCAL_DIR / "data" / "cda.db"
-PID_FILE = LOCAL_DIR / "run" / "watcher.pid"
-UI_PID_FILE = LOCAL_DIR / "run" / "ui.pid"
-UI_LOG_FILE = LOCAL_DIR / "logs" / "ui.log"
-WATCHER = PACKAGE_DIR.parent / "pipeline" / "watcher.py"
-INGEST = PACKAGE_DIR.parent / "pipeline" / "ingest.py"
-RECON = PACKAGE_DIR.parent / "pipeline" / "reconstruct.py"
-EXTRACT = PACKAGE_DIR.parent / "pipeline" / "extract.py"
-EMBED = PACKAGE_DIR.parent / "pipeline" / "embed.py"
+# Ensure runtime dirs exist on every CLI invocation
+ensure_dirs()
 
 kernel = PMFKernel()
 
@@ -333,14 +327,13 @@ def status():
         click.echo(f"  Start with: {bold('cda watch start')}")
 
     # Queue status
-    queue_dir = LOCAL_DIR / "queue"
-    if queue_dir.exists():
-        pending = len(list(queue_dir.glob("*.json")))
-        completed = len(list(queue_dir.glob("*.completed")))
+    if QUEUE_DIR.exists():
+        pending = len(list(QUEUE_DIR.glob("*.json")))
+        completed = len(list(QUEUE_DIR.glob("*.completed")))
         click.echo(f"  Queue: {pending} pending, {completed} completed")
         if pending > 0:
             # Show last pending operation
-            pending_files = sorted(queue_dir.glob("*.json"))
+            pending_files = sorted(QUEUE_DIR.glob("*.json"))
             if pending_files:
                 try:
                     data = json.loads(pending_files[-1].read_text())
@@ -546,7 +539,7 @@ def embed():
 def embed_build():
     """Build semantic embeddings and session intelligence."""
     click.echo(yellow("  Building semantic intelligence..."))
-    result = subprocess.run([sys.executable, str(EMBED)], capture_output=False)
+    result = subprocess.run([sys.executable, "-m", "cda.pipeline.embed"], capture_output=False)
     if result.returncode == 0:
         click.echo(green("  Embed build complete"))
     else:
@@ -754,7 +747,7 @@ def sync():
     errors = 0
 
     click.echo(yellow("  Running full ingest — this rewrites the DB..."))
-    result = subprocess.run([sys.executable, str(INGEST)], capture_output=False)
+    result = subprocess.run([sys.executable, "-m", "cda.pipeline.ingest"], capture_output=False)
     if result.returncode != 0:
         click.echo(red("  Ingest failed"))
         finish_run(run_id, stages_done, {}, errors=1, exit_code=1, notes="ingest failed")
@@ -763,7 +756,7 @@ def sync():
 
     click.echo(green("  Ingest complete"))
     click.echo(yellow("  Running reconstruction..."))
-    result = subprocess.run([sys.executable, str(RECON)], capture_output=False)
+    result = subprocess.run([sys.executable, "-m", "cda.pipeline.reconstruct"], capture_output=False)
     if result.returncode != 0:
         click.echo(red("  Reconstruction failed"))
         finish_run(run_id, stages_done, {}, errors=1, exit_code=1, notes="reconstruct failed")
@@ -772,7 +765,7 @@ def sync():
 
     click.echo(green("  Reconstruction complete"))
     click.echo(yellow("  Running analysis..."))
-    result = subprocess.run([sys.executable, str(EXTRACT)], capture_output=False)
+    result = subprocess.run([sys.executable, "-m", "cda.pipeline.extract"], capture_output=False)
     if result.returncode != 0:
         click.echo(red("  Analysis failed"))
         finish_run(run_id, stages_done, {}, errors=1, exit_code=1, notes="extract failed")
@@ -781,7 +774,7 @@ def sync():
 
     click.echo(green("  Analysis complete"))
     click.echo(yellow("  Running semantic intelligence..."))
-    result = subprocess.run([sys.executable, str(EMBED)], capture_output=False)
+    result = subprocess.run([sys.executable, "-m", "cda.pipeline.embed"], capture_output=False)
     if result.returncode != 0:
         click.echo(red("  Semantic intelligence failed"))
         errors += 1
@@ -809,7 +802,7 @@ def sync():
 def reconstruct():
     """Re-run session reconstruction and FTS rebuild only."""
     click.echo(yellow("  Reconstructing exchanges..."))
-    subprocess.run([sys.executable, str(RECON)], capture_output=False)
+    subprocess.run([sys.executable, "-m", "cda.pipeline.reconstruct"], capture_output=False)
     click.echo(green("  Done"))
 
 
@@ -1470,9 +1463,8 @@ def policy():
 def policy_allow(pattern):
     """Add an allow pattern for search results."""
     # For now, store in a simple text file
-    policy_file = LOCAL_DIR / "config" / "policy.txt"
     try:
-        with open(policy_file, "a") as f:
+        with open(POLICY_FILE, "a") as f:
             f.write(f"ALLOW {pattern}\n")
         click.echo(green(f"  Added allow pattern: {pattern}"))
     except Exception as e:
@@ -1483,9 +1475,8 @@ def policy_allow(pattern):
 @click.argument("pattern")
 def policy_deny(pattern):
     """Add a deny pattern for search results."""
-    policy_file = LOCAL_DIR / "config" / "policy.txt"
     try:
-        with open(policy_file, "a") as f:
+        with open(POLICY_FILE, "a") as f:
             f.write(f"DENY {pattern}\n")
         click.echo(green(f"  Added deny pattern: {pattern}"))
     except Exception as e:
@@ -1495,8 +1486,7 @@ def policy_deny(pattern):
 @policy.command("list")
 def policy_list():
     """List current policies."""
-    policy_file = LOCAL_DIR / "config" / "policy.txt"
-    if not policy_file.exists():
+    if not POLICY_FILE.exists():
         click.echo(dim("  No policies configured"))
         return
 
@@ -1504,7 +1494,7 @@ def policy_list():
     click.echo(bold("  Data Access Policies"))
     click.echo(hr())
     try:
-        with open(policy_file, "r") as f:
+        with open(POLICY_FILE, "r") as f:
             for line in f:
                 line = line.strip()
                 if line.startswith("ALLOW "):
@@ -1518,14 +1508,13 @@ def policy_list():
 
 def check_policy(text):
     """Check if text passes policy filters. Returns True if allowed."""
-    policy_file = LOCAL_DIR / "config" / "policy.txt"
-    if not policy_file.exists():
+    if not POLICY_FILE.exists():
         return True  # No policies = allow all
 
     allow_patterns = []
     deny_patterns = []
     try:
-        with open(policy_file, "r") as f:
+        with open(POLICY_FILE, "r") as f:
             for line in f:
                 line = line.strip()
                 if line.startswith("ALLOW "):
@@ -2574,6 +2563,55 @@ def check(as_json, fail_fast):
     sys.exit(0 if passed_all else 1)
 
 
+# ─────────────────────────────────────────────
+# INIT
+# ─────────────────────────────────────────────
+
+@cli.command("init")
+def init():
+    """First-run setup — create ~/.cda/ directory structure and validate environment."""
+    from cda.kernel.paths import (
+        CDA_HOME, DATA_DIR, RUN_DIR, LOG_DIR, QUEUE_DIR,
+        PMF_DIR, PMF_LOG_DIR, CONFIG_DIR, POLICY_FILE,
+    )
+    import os
+
+    click.echo()
+    click.echo(bold("  Code Data Ark — init"))
+    click.echo(hr())
+
+    # Create directory tree
+    dirs = [DATA_DIR, RUN_DIR, LOG_DIR, QUEUE_DIR, PMF_DIR, PMF_LOG_DIR, CONFIG_DIR]
+    for d in dirs:
+        d.mkdir(parents=True, exist_ok=True)
+        click.echo(green(f"  {d}"))
+
+    # Write a starter policy file if none exists
+    if not POLICY_FILE.exists():
+        POLICY_FILE.write_text("# CDA access policy\n# ALLOW <pattern>\n# DENY  <pattern>\n")
+        click.echo(green(f"  {POLICY_FILE}  (created)"))
+
+    # Validate VS Code data dir
+    vscode_data = Path(os.environ.get(
+        "VSCODE_DATA_DIR",
+        Path.home() / "Library/Application Support/Code/User",
+    ))
+    if vscode_data.exists():
+        click.echo(green(f"  VS Code data dir: {vscode_data}"))
+    else:
+        click.echo(yellow(f"  VS Code data dir not found: {vscode_data}"))
+        click.echo(yellow("  Set VSCODE_DATA_DIR if your data is elsewhere."))
+
+    click.echo()
+    click.echo(bold("  CDA_HOME: ") + str(CDA_HOME))
+    click.echo()
+    click.echo(dim("  Next steps:"))
+    click.echo(dim("    cda sync         — ingest all VS Code session data"))
+    click.echo(dim("    cda watch start  — start the live watcher daemon"))
+    click.echo(dim("    cda serve        — open the web dashboard on :10001"))
+    click.echo()
+
+
 # ─────────────────────────────────────────────
 # ENTRY
 # ─────────────────────────────────────────────

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/cda/ui/web.py

@@ -11,18 +11,14 @@ import threading
 import time
 import traceback
 import subprocess
+import sys
 import socket
 from typing import Any, Dict
-from pathlib import Path
 from datetime import datetime
 from wsgiref.simple_server import make_server, WSGIServer
 from urllib.parse import parse_qs
 from cda.kernel.pmf_kernel import PMFKernel
-
-# Get DB path relative to this file
-PACKAGE_DIR = Path(__file__).resolve().parent
-LOCAL_DIR = PACKAGE_DIR.parent.parent.parent / "local"
-DB_PATH = LOCAL_DIR / "data" / "cda.db"
+from cda.kernel.paths import DB_PATH
 kernel = PMFKernel()
 
 # ─────────────────────────────────────────────
@@ -1396,28 +1392,28 @@ def run_action_background(action_id, action_name):
     try:
         if action_name == "sync":
             result = subprocess.run(
-                ["python3", str(PACKAGE_DIR.parent / "pipeline" / "ingest.py")],
+                [sys.executable, "-m", "cda.pipeline.ingest"],
                 capture_output=True,
                 text=True,
                 timeout=300
             )
         elif action_name == "reconstruct":
             result = subprocess.run(
-                ["python3", str(PACKAGE_DIR.parent / "pipeline" / "reconstruct.py")],
+                [sys.executable, "-m", "cda.pipeline.reconstruct"],
                 capture_output=True,
                 text=True,
                 timeout=300
             )
         elif action_name == "embed-build":
             result = subprocess.run(
-                ["python3", str(PACKAGE_DIR.parent / "pipeline" / "embed.py"), "build"],
+                [sys.executable, "-m", "cda.pipeline.embed", "build"],
                 capture_output=True,
                 text=True,
                 timeout=600
             )
         elif action_name == "watch-start":
             result = subprocess.run(
-                ["python3", str(PACKAGE_DIR.parent / "pipeline" / "watcher.py"), "start"],
+                [sys.executable, "-m", "cda.pipeline.watcher", "start"],
                 capture_output=True,
                 text=True,
                 timeout=30

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/changelog.md

@@ -5,6 +5,21 @@ All notable changes to this project 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).
 
+## [2.0.3] - 2026-05-11
+
+### Fixed
+- **Install path resolution**: `LOCAL_DIR`/`DB_PATH` no longer derived from `__file__` — now resolves to `~/.cda/` (or `$CDA_HOME`). Survives `pip install` into site-packages.
+- All pipeline stages (`ingest`, `reconstruct`, `extract`, `embed`, `watcher`, `parse_edits`) import canonical paths from new `cda.kernel.paths` module.
+- `PMFKernel` and `selfcheck` updated to use `cda.kernel.paths`.
+- All subprocess pipeline invocations switched from script file paths to `python -m cda.pipeline.<stage>` module calls.
+
+### Added
+- `cda.kernel.paths` — single source of truth for `CDA_HOME`, `DB_PATH`, `PID_FILE`, `QUEUE_DIR`, `POLICY_FILE`, `ensure_dirs()`.
+- `cda init` command — first-run setup: creates `~/.cda/` directory tree, writes starter policy, validates VS Code data path.
+
+### Changed
+- README quickstart now reflects correct install flow: `pip install` → `cda init` → `cda sync` → `cda watch start` → `cda serve`.
+
 ## [2.0.2] - 2026-05-11
 
 ### Fixed

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "code-data-ark"
-version = "2.0.2"
+version = "2.0.3"
 description = "Code Data Ark — local observability and intelligence platform for VS Code + Copilot Chat sessions"
 readme = "readme.md"
 license = "MIT"

{code_data_ark-2.0.2 → code_data_ark-2.0.3}/readme.md

@@ -78,39 +78,41 @@ make install-dev
 
 ## ⚡ Quick Start
 
-1. **Initialize the database**
+1. **Install**
 
 ```bash
-cda sync
+pip install code-data-ark
 ```
 
-2. **Start the watcher daemon**
+2. **Initialize — create `~/.cda/` and validate your VS Code data path**
 
 ```bash
-cda watch start
+cda init
 ```
 
-3. **Inspect the PMF runtime services**
+3. **Ingest all VS Code session data**
 
 ```bash
-cda pmf services
+cda sync
 ```
 
-4. **Build semantic intelligence**
+4. **Start the live watcher daemon**
 
 ```bash
-cda embed build
+cda watch start
 ```
 
-4. **Start the web UI**
+5. **Open the web dashboard**
 
 ```bash
-cda ui start
+cda serve   # → http://127.0.0.1:10001
 ```
 
-5. **Open your browser**
+6. **Build semantic intelligence** (optional, requires `sentence-transformers`)
 
-Visit `http://127.0.0.1:10001`
+```bash
+cda embed build
+```
 
 ## 🌐 Web UI
 

code_data_ark-2.0.3/version

@@ -0,0 +1 @@
+2.0.3

code_data_ark-2.0.2/version

@@ -1 +0,0 @@
-2.0.2