ckgraphify 0.1.4__tar.gz → 0.2.0__tar.gz

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ckgraphify
-Version: 0.1.4
+Version: 0.2.0
 Summary: AI coding assistant skill for Claude Code and Codex - graph-main boundary graphs, multi-repo call chains, business-map concepts, and business search
 License-Expression: MIT
 Project-URL: Homepage, https://github.com/safishamsi/graphify
@@ -106,17 +106,17 @@ python3 -m pip install -e ./ckgraphify && CLAUDE_PLUGIN_ROOT=/Users/alsc/code/sh
 
 ## 生成并合并仓库图谱
 
-下面命令会分别进入 `${CLAUDE_PLUGIN_ROOT}/repos/kl-health` 和 `${CLAUDE_PLUGIN_ROOT}/repos/ele-newretail-health-audit` 生成 `graphify-out/graph-main.json`，然后合并到 `${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json`。
+下面命令会分别进入 `${CLAUDE_PLUGIN_ROOT}/repos/kl-health` 和 `${CLAUDE_PLUGIN_ROOT}/repos/ele-newretail-health-audit` 生成 `graphify-out/graph-main.json`，然后合并到 `${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json`。
 
 ```bash
-scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" kl-health ele-newretail-health-audit
+scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" kl-health ele-newretail-health-audit
 ```
 
 也可以传入更多 `${CLAUDE_PLUGIN_ROOT}/repos/` 下的仓库名：
 
 ```bash
 # 健康卡 init all repos
-scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" ele-newretail-drug ele-newretail-health-task ele-newretail-drug-grow ele-newretail-health-audit ele-newretail-health-client ele-newretail-drug-trade kl-health health-vip-card medicine-unicore p ele-newretail-venus ele-newretail-summaryx china-alsc-sales-eleme-newretail-app
+scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" ele-newretail-drug ele-newretail-health-task ele-newretail-drug-grow ele-newretail-health-audit ele-newretail-health-client ele-newretail-drug-trade kl-health health-vip-card medicine-unicore p ele-newretail-venus ele-newretail-summaryx china-alsc-sales-eleme-newretail-app
 ```
 
 ```bash
@@ -124,7 +124,7 @@ scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${
 scripts/graph-main-repos.sh \
     --incremental \
     --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" \
-    --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     p medicine-unicore
 ```
 
@@ -136,7 +136,7 @@ scripts/graph-main-repos.sh \
 关于ehealth-member实体的全部链路
 ```bash
 graphify main-trace \
-    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     --from ehealth-member \
     --max-depth 8
 ```
@@ -144,7 +144,7 @@ graphify main-trace \
 选择ehealth-member实体的/shopping/healthCard/createHealthCard 支线链路
 ```bash
 graphify main-trace \
-    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     --from ehealth-member \
     --api /shopping/healthCard/createHealthCard \
     --max-depth 8
@@ -154,7 +154,7 @@ graphify main-trace \
 
 ```bash
 graphify main-trace \
-    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     --from ehealth-member \
     --max-depth 8 --sources
 ```

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/README.md

@@ -13,17 +13,17 @@ python3 -m pip install -e ./ckgraphify && CLAUDE_PLUGIN_ROOT=/Users/alsc/code/sh
 
 ## 生成并合并仓库图谱
 
-下面命令会分别进入 `${CLAUDE_PLUGIN_ROOT}/repos/kl-health` 和 `${CLAUDE_PLUGIN_ROOT}/repos/ele-newretail-health-audit` 生成 `graphify-out/graph-main.json`，然后合并到 `${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json`。
+下面命令会分别进入 `${CLAUDE_PLUGIN_ROOT}/repos/kl-health` 和 `${CLAUDE_PLUGIN_ROOT}/repos/ele-newretail-health-audit` 生成 `graphify-out/graph-main.json`，然后合并到 `${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json`。
 
 ```bash
-scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" kl-health ele-newretail-health-audit
+scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" kl-health ele-newretail-health-audit
 ```
 
 也可以传入更多 `${CLAUDE_PLUGIN_ROOT}/repos/` 下的仓库名：
 
 ```bash
 # 健康卡 init all repos
-scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" ele-newretail-drug ele-newretail-health-task ele-newretail-drug-grow ele-newretail-health-audit ele-newretail-health-client ele-newretail-drug-trade kl-health health-vip-card medicine-unicore p ele-newretail-venus ele-newretail-summaryx china-alsc-sales-eleme-newretail-app
+scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" ele-newretail-drug ele-newretail-health-task ele-newretail-drug-grow ele-newretail-health-audit ele-newretail-health-client ele-newretail-drug-trade kl-health health-vip-card medicine-unicore p ele-newretail-venus ele-newretail-summaryx china-alsc-sales-eleme-newretail-app
 ```
 
 ```bash
@@ -31,7 +31,7 @@ scripts/graph-main-repos.sh --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" --out "${
 scripts/graph-main-repos.sh \
     --incremental \
     --repos-root "${CLAUDE_PLUGIN_ROOT}/repos" \
-    --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     p medicine-unicore
 ```
 
@@ -43,7 +43,7 @@ scripts/graph-main-repos.sh \
 关于ehealth-member实体的全部链路
 ```bash
 graphify main-trace \
-    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     --from ehealth-member \
     --max-depth 8
 ```
@@ -51,7 +51,7 @@ graphify main-trace \
 选择ehealth-member实体的/shopping/healthCard/createHealthCard 支线链路
 ```bash
 graphify main-trace \
-    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     --from ehealth-member \
     --api /shopping/healthCard/createHealthCard \
     --max-depth 8
@@ -61,7 +61,7 @@ graphify main-trace \
 
 ```bash
 graphify main-trace \
-    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json" \
+    --graph "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json" \
     --from ehealth-member \
     --max-depth 8 --sources
 ```

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/graphify/__main__.py

@@ -1330,9 +1330,9 @@ def main() -> None:
                 sys.exit(1)
             else:
                 graph_paths.append(Path(a)); i += 1
-        if len(graph_paths) < 2:
+        if len(graph_paths) < 1:
             print(
-                "Usage: graphify merge-main-graphs <graph-main1.json> <graph-main2.json> [...] "
+                "Usage: graphify merge-main-graphs <graph-main1.json> [graph-main2.json ...] "
                 "[--out graph-main-merged.json] [--report report.md]",
                 file=sys.stderr,
             )

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/graphify/business_map.py

@@ -345,9 +345,12 @@ def default_business_graph_path(map_path: Path, start: Path | None = None) -> Pa
     plugin_root = os.environ.get("CLAUDE_PLUGIN_ROOT", "").strip()
     if plugin_root:
         graphify_out = Path(plugin_root).expanduser().resolve() / "graphify-out"
-        hc_graph = graphify_out / "graph-hc.json"
-        if hc_graph.exists():
-            return hc_graph
+        graph = graphify_out / "graph.json"
+        if graph.exists():
+            return graph
+        legacy_graph = graphify_out / "graph-hc.json"
+        if legacy_graph.exists():
+            return legacy_graph
         candidates = sorted(
             p for p in graphify_out.glob("graph*.json")
             if p.name != "business-map.json"

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/graphify/graph_main_frontend.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 import json
 import os
 import re
+import hashlib
 from collections import deque
 from dataclasses import dataclass
 from datetime import datetime, timezone
@@ -12,6 +13,7 @@ from urllib.parse import urlparse
 from graphify.graph_main_frontend_sdk import (
     collect_sdk_dependencies,
     collect_sdk_exports,
+    collect_sdk_internal_links,
     detect_frontend_sdk_repo,
 )
 from graphify.repo_registry import repo_metadata_for_root
@@ -1132,6 +1134,13 @@ def build_graph_main_frontend(*, repo_root: Path, out_path: Path) -> GraphMainFr
         edge_ids.add(k)
         edges.append(e)
 
+    def unique_node_id(prefix: str, suffix: str, canonical: str) -> str:
+        base = f"{prefix}__{_safe_id(suffix)}"
+        if base not in node_ids:
+            return base
+        digest = hashlib.sha1(canonical.encode("utf-8")).hexdigest()[:8]
+        return f"{base}_{digest}"
+
     page_by_key: dict[str, str] = {}
     component_by_key: dict[str, str] = {}
     mtop_by_key: dict[str, str] = {}
@@ -1232,7 +1241,7 @@ def build_graph_main_frontend(*, repo_root: Path, out_path: Path) -> GraphMainFr
         if canonical in sdk_export_by_key:
             return sdk_export_by_key[canonical]
         suffix = canonical.split(":", 1)[-1]
-        nid = f"sdk_export__{_safe_id(suffix)}"
+        nid = unique_node_id("sdk_export", suffix, canonical)
         add_node(
             {
                 "id": nid,
@@ -1266,7 +1275,7 @@ def build_graph_main_frontend(*, repo_root: Path, out_path: Path) -> GraphMainFr
         if canonical in sdk_dependency_by_key:
             return sdk_dependency_by_key[canonical]
         suffix = canonical.split(":", 1)[-1]
-        nid = f"sdk_dependency__{_safe_id(suffix)}"
+        nid = unique_node_id("sdk_dependency", suffix, canonical)
         add_node(
             {
                 "id": nid,
@@ -1768,7 +1777,8 @@ def build_graph_main_frontend(*, repo_root: Path, out_path: Path) -> GraphMainFr
         mode = "frontend-sdk"
         detection["mode"] = mode
         detection["sdk_detection"] = sdk_detection.as_dict()
-        for export in collect_sdk_exports(repo_root):
+        sdk_exports = collect_sdk_exports(repo_root)
+        for export in sdk_exports:
             ensure_sdk_export(
                 kind=export.kind,
                 label=export.label,
@@ -1779,6 +1789,25 @@ def build_graph_main_frontend(*, repo_root: Path, out_path: Path) -> GraphMainFr
                 source_file=export.source_file,
                 source_location=export.source_location,
             )
+        for link in collect_sdk_internal_links(repo_root, sdk_exports):
+            source_nid = sdk_export_by_key.get(link.source_key)
+            target_nid = sdk_export_by_key.get(link.target_key)
+            if not source_nid or not target_nid:
+                continue
+            add_edge(
+                {
+                    "source": source_nid,
+                    "target": target_nid,
+                    "relation": link.relation,
+                    "confidence": "EXTRACTED",
+                    "confidence_score": 1.0,
+                    "source_file": link.source_file,
+                    "source_location": link.source_location,
+                    "weight": 1.0,
+                    "evidence": link.evidence,
+                    **({"resolution_kind": link.resolution_kind} if link.resolution_kind else {}),
+                }
+            )
     else:
         detection["sdk_detection"] = sdk_detection.as_dict()
         propagated_file_owners = propagate_file_owners(max_depth=3)

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/graphify/graph_main_frontend_sdk.py

@@ -111,6 +111,17 @@ class SdkDependency:
         return base
 
 
+@dataclass(frozen=True)
+class SdkInternalLink:
+    source_key: str
+    target_key: str
+    relation: str
+    source_file: str
+    source_location: str = "L1"
+    evidence: str = ""
+    resolution_kind: str = ""
+
+
 def _read_json(path: Path) -> dict:
     try:
         return json.loads(path.read_text(encoding="utf-8"))
@@ -256,17 +267,45 @@ def _strip_known_ext(rel: str) -> str:
     return rel
 
 
-def _export_path_for_rel(rel: str) -> str:
-    if rel.startswith("src/"):
-        rel = rel[len("src/") :]
+def _export_path_for_rel(rel: str, package_root: str = "src") -> str:
+    prefix = f"{package_root.strip('/')}/"
+    if package_root.strip("/") == "src" and rel.startswith(prefix):
+        rel = rel[len(prefix) :]
     return _strip_known_ext(rel)
 
 
+def _package_source_roots(repo_root: Path) -> list[Path]:
+    roots: list[Path] = []
+    pkg = package_json(repo_root)
+    main = str(pkg.get("main", "") or "").strip()
+    if main:
+        first = main.split("/", 1)[0]
+        if first and (repo_root / first).is_dir():
+            roots.append(repo_root / first)
+    for name in ("src", "lib"):
+        p = repo_root / name
+        if p.is_dir() and p not in roots:
+            roots.append(p)
+    return roots
+
+
+def _source_root_name(repo_root: Path, path: Path) -> str:
+    try:
+        rel = path.relative_to(repo_root)
+    except ValueError:
+        return ""
+    return rel.parts[0] if rel.parts else ""
+
+
 def _is_component_entry(rel: str) -> bool:
-    if not rel.startswith("src/components/"):
+    if rel.startswith("src/"):
+        rel = rel[len("src/") :]
+    elif rel.startswith("lib/"):
+        rel = rel[len("lib/") :]
+    if not rel.startswith("components/") and not rel.startswith("component/"):
         return False
     parts = rel.split("/")
-    if len(parts) != 5:
+    if len(parts) < 4:
         return False
     if parts[-1].split(".", 1)[0] != "index":
         return False
@@ -274,7 +313,49 @@ def _is_component_entry(rel: str) -> bool:
 
 
 def _is_sdk_extension_file(rel: str) -> bool:
-    return rel.startswith("src/extensions/")
+    if rel.startswith("src/"):
+        rel = rel[len("src/") :]
+    elif rel.startswith("lib/"):
+        rel = rel[len("lib/") :]
+    return rel.startswith("extensions/")
+
+
+def _strip_package_root(rel: str) -> str:
+    if rel.startswith("src/"):
+        return rel[len("src/") :]
+    if rel.startswith("lib/"):
+        return rel[len("lib/") :]
+    return rel
+
+
+def _is_publishable_sdk_file(rel: str) -> bool:
+    rel = _strip_package_root(rel)
+    basename = Path(rel).name
+    if basename.startswith("index."):
+        return True
+    if rel.startswith("extensions/"):
+        return True
+    if "/Hooks/" in rel or "/hooks/" in rel:
+        return True
+    if "/component/" in f"/{rel}" or "/components/" in f"/{rel}":
+        return True
+    if "/core/" in f"/{rel}" or "/util/" in f"/{rel}":
+        return True
+    return False
+
+
+def _sdk_kind_for_rel(rel: str, export_path: str, export_name: str = "") -> str:
+    stripped = _strip_package_root(rel)
+    lower = stripped.lower()
+    if "/hooks/" in lower or "/hook/" in lower:
+        return "hook"
+    if _is_component_entry(rel) or "/component/" in f"/{lower}" or "/components/" in f"/{lower}":
+        return "component" if not export_name or export_name[:1].isupper() else "function"
+    if _is_sdk_extension_file(rel):
+        return "function"
+    if export_path.startswith("core/"):
+        return "function" if export_name else "module"
+    return "function" if export_name else "module"
 
 
 def _line_of(text: str, idx: int) -> str:
@@ -304,6 +385,26 @@ def _named_exports(text: str) -> list[tuple[str, str]]:
     return sorted(dedup.items())
 
 
+def _root_reexport_names(text: str) -> list[tuple[str, str]]:
+    out: list[tuple[str, str]] = []
+    for m in _NAMED_EXPORT_RE.finditer(text):
+        body = m.group("body")
+        for part in body.split(","):
+            token = part.strip()
+            if not token:
+                continue
+            if " as " in token:
+                token = token.split(" as ", 1)[1].strip()
+            else:
+                token = token.split(" ", 1)[0].strip()
+            if token:
+                out.append((token, _line_of(text, m.start())))
+    dedup: dict[str, str] = {}
+    for name, line in out:
+        dedup.setdefault(name, line)
+    return sorted(dedup.items())
+
+
 def _has_default_export(text: str) -> tuple[bool, str]:
     m = _EXPORT_DEFAULT_RE.search(text)
     if not m:
@@ -335,24 +436,48 @@ def collect_sdk_exports(repo_root: Path) -> list[SdkExport]:
     def add(item: SdkExport) -> None:
         exports[(item.kind, item.export_path, item.export_name)] = item
 
-    for path in _iter_code_files(repo_root / "src"):
-        rel = _norm_rel(repo_root, path)
-        export_path = _export_path_for_rel(rel)
-        text = _read_text(path)
-        default_export, default_line = _has_default_export(text)
-        default_name = _default_export_name(text, rel) if default_export else ""
-        named_exports = _named_exports(text)
-
-        if _is_component_entry(rel):
-            add(SdkExport("component", pkg_name, export_path, "", rel, default_line or "L1"))
-
-        if _is_sdk_extension_file(rel):
+    for source_root in _package_source_roots(repo_root):
+        root_name = _source_root_name(repo_root, source_root)
+        if not root_name:
+            continue
+        for path in _iter_code_files(source_root):
+            rel = _norm_rel(repo_root, path)
+            if not _is_publishable_sdk_file(rel):
+                continue
+            export_path = _export_path_for_rel(rel, root_name)
+            text = _read_text(path)
+            default_export, default_line = _has_default_export(text)
+            default_name = _default_export_name(text, rel) if default_export else ""
+            named_exports = _named_exports(text)
+
+            if _strip_known_ext(_strip_package_root(rel)) == "index":
+                if default_export:
+                    add(SdkExport("module", pkg_name, "", "default", rel, default_line or "L1"))
+                for name, line in _root_reexport_names(text):
+                    kind = "component" if name[:1].isupper() else "function"
+                    add(SdkExport(kind, pkg_name, "", name, rel, line))
+
+            if _is_component_entry(rel):
+                add(SdkExport("component", pkg_name, export_path, "", rel, default_line or "L1"))
+
+            if _is_sdk_extension_file(rel):
+                for name, line in named_exports:
+                    kind = "hook" if "/hooks/" in rel.lower() else "function"
+                    add(SdkExport(kind, pkg_name, export_path, name, rel, line))
+                if default_export:
+                    kind = "hook" if "/hooks/" in rel.lower() else "function"
+                    add(SdkExport(kind, pkg_name, export_path, default_name, rel, default_line))
+
+            if default_export and export_path:
+                kind = _sdk_kind_for_rel(rel, export_path, default_name)
+                export_name = default_name if kind != "component" else ""
+                add(SdkExport(kind, pkg_name, export_path, export_name, rel, default_line or "L1"))
+                file_export_name = Path(_strip_known_ext(export_path)).name
+                if kind != "component" and file_export_name and file_export_name != export_name:
+                    add(SdkExport(kind, pkg_name, export_path, file_export_name, rel, default_line or "L1"))
             for name, line in named_exports:
-                kind = "hook" if "/hooks/" in rel else "function"
+                kind = _sdk_kind_for_rel(rel, export_path, name)
                 add(SdkExport(kind, pkg_name, export_path, name, rel, line))
-            if default_export:
-                kind = "hook" if "/hooks/" in rel else "function"
-                add(SdkExport(kind, pkg_name, export_path, default_name, rel, default_line))
 
     return sorted(exports.values(), key=lambda e: (e.kind, e.export_path, e.export_name))
 
@@ -438,6 +563,170 @@ def _sdk_member_uses(text: str, alias: str) -> set[str]:
     return {m.group(1) for m in pat.finditer(text)}
 
 
+def _resolve_sdk_relative_import(repo_root: Path, current_file: Path, import_path: str) -> Path | None:
+    imp = (import_path or "").strip()
+    if not imp.startswith("."):
+        return None
+    base = (current_file.parent / imp).resolve()
+    candidates = [
+        base,
+        base.with_suffix(".ts"),
+        base.with_suffix(".tsx"),
+        base.with_suffix(".js"),
+        base.with_suffix(".jsx"),
+        base / "index.ts",
+        base / "index.tsx",
+        base / "index.js",
+        base / "index.jsx",
+    ]
+    try:
+        resolved_root = repo_root.resolve()
+    except OSError:
+        return None
+    for candidate in candidates:
+        if not candidate.exists() or not candidate.is_file():
+            continue
+        try:
+            candidate.resolve().relative_to(resolved_root)
+        except ValueError:
+            continue
+        return candidate
+    return None
+
+
+def _export_basename(export: SdkExport) -> str:
+    raw = export.export_path.rstrip("/")
+    return raw.rsplit("/", 1)[-1] if raw else ""
+
+
+def _target_exports_for_import(
+    exports: list[SdkExport],
+    *,
+    default_alias: str = "",
+    named_name: str = "",
+) -> list[SdkExport]:
+    if not exports:
+        return []
+    if named_name:
+        wanted = named_name.lower()
+        return [item for item in exports if item.export_name.lower() == wanted]
+    alias = default_alias.lower()
+    if alias:
+        matched = [
+            item
+            for item in exports
+            if item.export_name.lower() == alias or _export_basename(item).lower() == alias
+        ]
+        if matched:
+            return matched
+    return [item for item in exports if item.export_name in {"", "default"}] or exports[:1]
+
+
+def _source_exports_for_alias(exports: list[SdkExport], alias: str) -> list[SdkExport]:
+    if not alias:
+        return exports
+    wanted = alias.lower()
+    return [
+        item
+        for item in exports
+        if item.export_name.lower() == wanted or _export_basename(item).lower() == wanted
+    ]
+
+
+def _line_for_text(text: str, token: str) -> str:
+    idx = text.find(token)
+    return _line_of(text, idx if idx >= 0 else 0)
+
+
+def _alias_is_used(text: str, alias: str) -> bool:
+    if not alias:
+        return False
+    return bool(re.search(rf"""\b{re.escape(alias)}\b""", text))
+
+
+def collect_sdk_internal_links(repo_root: Path, exports: list[SdkExport]) -> list[SdkInternalLink]:
+    by_file: dict[str, list[SdkExport]] = {}
+    for export in exports:
+        by_file.setdefault(export.source_file, []).append(export)
+
+    links: dict[tuple[str, str, str], SdkInternalLink] = {}
+
+    def add(link: SdkInternalLink) -> None:
+        if link.source_key == link.target_key:
+            return
+        links[(link.source_key, link.target_key, link.relation)] = link
+
+    for rel, source_exports in sorted(by_file.items()):
+        source_path = repo_root / rel
+        text = _read_text(source_path)
+        if not text:
+            continue
+        stripped = _strip_js_comments(text)
+        import_aliases: dict[str, list[SdkExport]] = {}
+
+        for m in _IMPORT_FROM_RE.finditer(stripped):
+            raw_import = m.group("path")
+            target_path = _resolve_sdk_relative_import(repo_root, source_path, raw_import)
+            if not target_path:
+                continue
+            target_exports = by_file.get(_norm_rel(repo_root, target_path), [])
+            if not target_exports:
+                continue
+            default_alias, named = _parse_import_clause(m.group("clause"))
+            if default_alias:
+                import_aliases[default_alias] = _target_exports_for_import(target_exports, default_alias=default_alias)
+            for name in named:
+                import_aliases[name] = _target_exports_for_import(target_exports, named_name=name)
+
+        exported_aliases = {name for name, _ in _root_reexport_names(stripped)}
+        for alias, targets in import_aliases.items():
+            if not targets:
+                continue
+            relation = "resolves_sdk_export" if alias in exported_aliases else "uses_sdk_export"
+            if relation == "uses_sdk_export" and not _alias_is_used(stripped, alias):
+                continue
+            source_candidates = _source_exports_for_alias(source_exports, alias) if relation == "resolves_sdk_export" else source_exports
+            for source in source_candidates:
+                for target in targets:
+                    add(
+                        SdkInternalLink(
+                            source_key=source.npm_export_key,
+                            target_key=target.npm_export_key,
+                            relation=relation,
+                            source_file=rel,
+                            source_location=_line_for_text(text, alias),
+                            evidence="sdk_relative_import",
+                            resolution_kind="reexport" if relation == "resolves_sdk_export" else "import_usage",
+                        )
+                    )
+
+        for m in _EXPORT_FROM_RE.finditer(stripped):
+            raw_import = m.group("path")
+            target_path = _resolve_sdk_relative_import(repo_root, source_path, raw_import)
+            if not target_path:
+                continue
+            target_exports = by_file.get(_norm_rel(repo_root, target_path), [])
+            if not target_exports:
+                continue
+            default_alias, named = _parse_import_clause(m.group("clause"))
+            for alias in named or ([default_alias] if default_alias else []):
+                for source in _source_exports_for_alias(source_exports, alias):
+                    for target in _target_exports_for_import(target_exports, named_name=alias):
+                        add(
+                            SdkInternalLink(
+                                source_key=source.npm_export_key,
+                                target_key=target.npm_export_key,
+                                relation="resolves_sdk_export",
+                                source_file=rel,
+                                source_location=_line_for_text(text, raw_import),
+                                evidence="sdk_export_from_relative_import",
+                                resolution_kind="export_from",
+                            )
+                        )
+
+    return sorted(links.values(), key=lambda item: (item.source_key, item.relation, item.target_key))
+
+
 def collect_sdk_dependencies(repo_root: Path, code_files: list[Path]) -> list[SdkDependency]:
     packages = set(sdk_package_allowlist(repo_root).keys())
     if not packages:

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/graphify/graph_main_merge.py

@@ -8,7 +8,7 @@ from datetime import datetime, timezone
 from pathlib import Path
 from typing import Iterable
 
-from graphify.repo_registry import repo_metadata_by_name
+from graphify.repo_registry import repo_id_for, repo_metadata_by_name
 
 
 _ANCHOR_KINDS = {"mtop_api", "rest_api", "hsf_api", "hsf_method", "metaq_producer", "metaq_consumer"}
@@ -49,20 +49,24 @@ def _repo_meta_from_graph(data: dict, fallback_name: str) -> dict:
     raw = graph.get("repo", {}) if isinstance(graph, dict) else {}
     if isinstance(raw, dict):
         name = str(raw.get("name", "") or graph.get("repo_name", "") or fallback_name)
+        group = str(raw.get("group", "") or "")
+        repo_id = str(raw.get("id", "") or raw.get("repo_id", "") or repo_id_for(group, name))
         return {
+            "id": repo_id,
             "name": name,
-            "group": str(raw.get("group", "") or ""),
+            "group": group,
             "git": str(raw.get("git", "") or ""),
         }
-    return {"name": str(graph.get("repo_name", "") or fallback_name), "group": "", "git": ""}
+    name = str(graph.get("repo_name", "") or fallback_name)
+    return {"id": repo_id_for("", name), "name": name, "group": "", "git": ""}
 
 
 def _norm_key(v: object) -> str:
     return str(v or "").strip().lower()
 
 
-def _node_key(repo: str, node_id: object) -> str:
-    return f"{repo}::{node_id}"
+def _node_key(repo_id: str, node_id: object) -> str:
+    return f"{repo_id}::{node_id}"
 
 
 def _shared_node_id(anchor: str) -> str:
@@ -78,21 +82,22 @@ def _edge_key(edge: dict) -> tuple[str, str, str, str]:
     )
 
 
-def _copy_node(repo: str, node: dict) -> dict:
+def _copy_node(repo_id: str, repo_name: str, node: dict) -> dict:
     out = dict(node)
     local_id = str(node.get("id", ""))
-    out["id"] = _node_key(repo, local_id)
+    out["id"] = _node_key(repo_id, local_id)
     out["local_id"] = local_id
-    out["repo"] = repo
+    out["repo"] = repo_id
+    out["repo_name"] = repo_name
     out.setdefault("main_kind", "unknown")
     return out
 
 
-def _copy_edge(repo: str, edge: dict) -> dict:
+def _copy_edge(repo_id: str, edge: dict) -> dict:
     out = dict(edge)
-    out["source"] = _node_key(repo, edge.get("source", ""))
-    out["target"] = _node_key(repo, edge.get("target", ""))
-    out["repo"] = repo
+    out["source"] = _node_key(repo_id, edge.get("source", ""))
+    out["target"] = _node_key(repo_id, edge.get("target", ""))
+    out["repo"] = repo_id
     out["edge_scope"] = "intra_repo"
     return out
 
@@ -197,8 +202,8 @@ def _load_graph(path: Path) -> dict:
 
 def merge_graph_main_files(graph_paths: Iterable[Path], out_path: Path) -> GraphMainMergeStats:
     inputs = [Path(p).resolve() for p in graph_paths]
-    if len(inputs) < 2:
-        raise ValueError("merge_graph_main_files requires at least two graph-main.json files")
+    if len(inputs) < 1:
+        raise ValueError("merge_graph_main_files requires at least one graph-main.json file")
 
     nodes: list[dict] = []
     edges: list[dict] = []
@@ -220,17 +225,21 @@ def merge_graph_main_files(graph_paths: Iterable[Path], out_path: Path) -> Graph
         data = _load_graph(path)
         path_repo = _repo_from_path(path)
         graph_repo_meta = _repo_meta_from_graph(data, path_repo)
-        repo = str(graph_repo_meta.get("name", "") or path_repo)
-        registry_meta = registry.get(repo, {})
+        repo_name = str(graph_repo_meta.get("name", "") or path_repo)
+        registry_meta = registry.get(repo_name, {})
         git = str(graph_repo_meta.get("git", "") or registry_meta.get("git", "") or "")
         group = str(graph_repo_meta.get("group", "") or registry_meta.get("group", "") or "")
+        branch = str(graph_repo_meta.get("branch", "") or registry_meta.get("branch", "") or "master")
+        repo_id = str(graph_repo_meta.get("id", "") or registry_meta.get("id", "") or repo_id_for(group, repo_name))
         repo_root_abs = path.parent.parent
-        repo_roots[repo] = f"repos/{repo}"
+        repo_roots[repo_id] = f"repos/{repo_name}"
         repo_meta = {
-            "repo": repo,
-            "name": repo,
+            "repo": repo_id,
+            "id": repo_id,
+            "name": repo_name,
             "group": group,
             "git": git,
+            "branch": branch,
             "kind": data.get("graph", {}).get("kind", ""),
             "mode": data.get("graph", {}).get("mode", ""),
             "nodes": len(data.get("nodes", [])),
@@ -239,7 +248,7 @@ def merge_graph_main_files(graph_paths: Iterable[Path], out_path: Path) -> Graph
         repos.append(repo_meta)
 
         for node in data.get("nodes", []):
-            copied = _copy_node(repo, node)
+            copied = _copy_node(repo_id, repo_name, node)
             nid = str(copied.get("id", ""))
             nodes.append(copied)
             id_to_node[nid] = copied
@@ -258,7 +267,7 @@ def merge_graph_main_files(graph_paths: Iterable[Path], out_path: Path) -> Graph
                 sdk_dependency_index[match_key].append(nid)
 
         for edge in data.get("links", []):
-            copied = _copy_edge(repo, edge)
+            copied = _copy_edge(repo_id, edge)
             key = _edge_key(copied)
             if key in edge_seen:
                 continue

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/graphify/graph_main_trace.py

@@ -20,6 +20,10 @@ _TRACE_RELS = {
     "resolves_rest_api",
     "resolves_hsf_api",
     "resolves_hsf_method",
+    "uses_sdk_dependency",
+    "resolves_npm_export",
+    "resolves_sdk_export",
+    "uses_sdk_export",
     "produces_metaq",
     "consumes_metaq",
     "resolves_metaq_topic",
@@ -34,6 +38,8 @@ _START_KINDS = {
     "mtop_api",
     "dependency",
     "dependency_method",
+    "sdk_dependency",
+    "sdk_export",
     "metaq_consumer",
     "metaq_producer",
 }
@@ -45,6 +51,8 @@ _TARGET_KINDS = {
     "hsf_method",
     "dependency",
     "dependency_method",
+    "sdk_dependency",
+    "sdk_export",
     "metaq_consumer",
     "metaq_producer",
 }

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/graphify/repo_registry.py

@@ -19,9 +19,26 @@ class RepoMetadata:
     name: str
     group: str = ""
     git: str = ""
+    branch: str = "master"
+    id: str = ""
 
     def as_dict(self) -> dict:
-        return {"name": self.name, "group": self.group, "git": self.git}
+        repo_id = self.id or repo_id_for(self.group, self.name)
+        return {"id": repo_id, "name": self.name, "group": self.group, "git": self.git, "branch": self.branch or "master"}
+
+
+def repo_id_for(group: str, name: str) -> str:
+    group = str(group or "").strip().strip("/")
+    name = str(name or "").strip().strip("/")
+    return f"{group}/{name}" if group else name
+
+
+def _split_repo_id(repo_id: str) -> tuple[str, str]:
+    raw = str(repo_id or "").strip().strip("/")
+    if "/" not in raw:
+        return "", raw
+    group, name = raw.rsplit("/", 1)
+    return group, name
 
 
 def resolve_repo_list_path(start: Path | None = None) -> Path:
@@ -97,14 +114,19 @@ def _read_repo_list(path: Path) -> list[RepoMetadata]:
     for raw in raw_entries:
         if not isinstance(raw, dict):
             continue
-        name = str(raw.get("name", "") or raw.get("repo", "")).strip()
+        raw_id = str(raw.get("id", "") or raw.get("repo_id", "")).strip()
+        id_group, id_name = _split_repo_id(raw_id)
+        name = str(raw.get("name", "") or raw.get("repo", "") or id_name).strip()
         if not name:
             continue
+        group = str(raw.get("group", "") or id_group or "")
         repos.append(
             RepoMetadata(
                 name=name,
-                group=str(raw.get("group", "") or ""),
+                group=group,
                 git=str(raw.get("git", "") or raw.get("url", "") or ""),
+                branch=str(raw.get("branch", "") or raw.get("default_branch", "") or raw.get("main_branch", "") or "master"),
+                id=raw_id or repo_id_for(group, name),
             )
         )
     return repos
@@ -112,10 +134,13 @@ def _read_repo_list(path: Path) -> list[RepoMetadata]:
 
 def _write_repo_list(path: Path, repos: Iterable[RepoMetadata]) -> None:
     path.parent.mkdir(parents=True, exist_ok=True)
-    ordered = sorted({r.name: r for r in repos}.values(), key=lambda r: (r.group, r.name))
+    ordered = sorted(
+        {(r.id or repo_id_for(r.group, r.name)): r for r in repos}.values(),
+        key=lambda r: (r.group, r.name),
+    )
     data = {
         "version": 1,
-        "description": "Editable ckgraphify repo registry. Edit name/group/git here; graph-main generation preserves non-empty manual values.",
+        "description": "Editable ckgraphify repo registry. Edit name/group/git/branch here; graph-main generation preserves non-empty manual values.",
         "repos": [r.as_dict() for r in ordered],
     }
     path.write_text(json.dumps(data, indent=2, ensure_ascii=False) + "\n", encoding="utf-8")
@@ -186,10 +211,13 @@ def ensure_repo_list(start: Path | None = None) -> Path:
                 by_name[scanned.name] = scanned
                 changed = True
             elif not current.git and scanned.git:
+                group = current.group or scanned.group
                 by_name[scanned.name] = RepoMetadata(
                     name=current.name,
-                    group=current.group or scanned.group,
+                    group=group,
                     git=scanned.git,
+                    branch=current.branch,
+                    id=repo_id_for(group, current.name),
                 )
                 changed = True
             elif not current.group and scanned.group:
@@ -197,6 +225,8 @@ def ensure_repo_list(start: Path | None = None) -> Path:
                     name=current.name,
                     group=scanned.group,
                     git=current.git,
+                    branch=current.branch,
+                    id=repo_id_for(scanned.group, current.name),
                 )
                 changed = True
 

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "ckgraphify"
-version = "0.1.4"
+version = "0.2.0"
 description = "AI coding assistant skill for Claude Code and Codex - graph-main boundary graphs, multi-repo call chains, business-map concepts, and business search"
 readme = "README.md"
 license = "MIT"

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/skill/skill-codex.md

@@ -73,7 +73,7 @@ Show this API briefly.
   - If the current directory is the parent of a marked `kg-banks/` root, use `./kg-banks/graphify-out/business-map.json`.
   - A marked `kg-banks/` root must contain `__root__` or `graphify-out/business-map.json`; do not trust the directory name alone.
 - Default graph: use `--graph`, then a JSON passed via `--path`, then the graph bound in `business-map.json` (`graph.path`), then `<resolved-map-root>/graphify-out/graph*.json`, then `graphify-out/graph-main-merged.json`.
-- For this kg-banks workspace, the bound default graph is `graphify-out/graph-hc.json`; from the parent directory it is `./kg-banks/graphify-out/graph-hc.json`.
+- For this kg-banks workspace, the bound default graph is `graphify-out/graph.json`; from the parent directory it is `./kg-banks/graphify-out/graph.json`.
 - Default depth: `10`.
 - Default answer: concise business result first, evidence second, gaps last.
 - If required inputs are missing and cannot be inferred from files, ask one short clarification.
@@ -97,7 +97,7 @@ Prefer existing deterministic commands:
 
 Treat the merged graph as business navigation, not as a complete execution trace. Its normal responsibility is to get from business wording to the right repo, page, MTop/REST API, HSF method, dependency boundary, and source filepath.
 
-Do not expect `graph-hc.json` or other merged graphs to describe every parameter, response field, extension point, or method-local branch. That level of detail would make the graph too large and noisy. When a question depends on field-level or parameter-level behavior, first use `main-trace` and source filepaths to land on the right code, then read the relevant implementation directly.
+Do not expect `graph.json` or other merged graphs to describe every parameter, response field, extension point, or method-local branch. That level of detail would make the graph too large and noisy. When a question depends on field-level or parameter-level behavior, first use `main-trace` and source filepaths to land on the right code, then read the relevant implementation directly.
 
 Missing field-level graph evidence is not automatically a graph defect. Classify it as one of:
 
@@ -126,7 +126,7 @@ Search discipline:
 For `$ckgraphify learn`:
 
 1. Require concept, scenario, repos, graph, map, and out.
-2. Read `ckgraphify/docs/business-map-policy.md`; every map iteration must follow it or explicitly report the conflict.
+2. Read the task-local `input/business-map-policy.md` when present; otherwise read `ckgraphify/docs/business-map-policy.md`. Every map iteration must follow it or explicitly report the conflict.
 3. Read existing map scenarios, flows, gaps, accepted gaps, verification scopes, and `search_constraints`.
 4. Search graph nodes with aliases, keywords, page/API names, HSF names, and repo hints.
 5. Run `main-trace` on promising graph matches.
@@ -153,7 +153,7 @@ Keep graph-main schema details, dependency extraction rules, and low-level branc
 ## Guardrails
 
 - Do not write business semantics into `graph-main.json`.
-- Before changing `business-map.json`, read `ckgraphify/docs/business-map-policy.md`; follow it or state the conflict.
+- Before changing `business-map.json`, read the task-local `input/business-map-policy.md` when present; otherwise read `ckgraphify/docs/business-map-policy.md`. Follow it or state the conflict.
 - Use `scenario.flows[]` for sub-links inside one complex business scene, such as landing page purchase/refund/benefit branches.
 - Keep `business-map.json` as a search/trace index. Do not add long `evidence` arrays there; put audit proof in `graphify-out/business-map-evidence.json`.
 - Use `gaps` for active unknowns and `accepted_gaps` for known boundaries with a stop rule.

{ckgraphify-0.1.4 → ckgraphify-0.2.0}/skill/skill.md

@@ -15,16 +15,12 @@ Before running any `graphify` command in the Claude plugin environment, ensure t
 export GRAPHIFY="${CLAUDE_PLUGIN_ROOT}/.graphify/bin/graphify"
 ```
 
+The setup script also runs `${CLAUDE_PLUGIN_ROOT}/scripts/sync-artifacts.sh`. Before using the default graph or business map, this sync script queries the production kg-server (`KG_SERVER_URL`, default `https://kg-banks.alibaba.net`) for the lightweight latest manifest. If the remote graph/concept versions and checksums already match the local manifest, it does not download files.
+
 If `${CLAUDE_PLUGIN_ROOT}` is missing, report that the Claude plugin environment is not active. Do not use macOS `/usr/bin/python3` or `sudo pip`. Do not inline the setup script from this skill; run the bundled script so quoting, checksum validation, and Python selection stay deterministic.
 
 ## User API
 
-```bash
-/ckgraphify init --path <workspace-or-repos-root> --concept <business-concept> [--graph <merged-graph.json>] [--map <business-map.json>]
-```
-
-Prepare the environment: locate or build repo `graph-main.json`, merge graphs if needed, initialize `business-map.json`, and validate anchors.
-
 ```bash
 /ckgraphify query "<business question>" --path <workspace-or-graph.json> [--map <business-map.json>] [--sources]
 ```
@@ -37,19 +33,6 @@ Semantic business search. First try to hit a known business concept/scenario/flo
 
 Trace a known concept or scenario. If no scenario is supplied, trace all useful known branches unless the result would be too broad.
 
-```bash
-/ckgraphify learn "<concept:scenario>" --path <workspace-or-repos-root> --graph <merged-graph.json> --map <business-map.json> --out <candidate.json> [--keywords "..."] [--docs <paths...>]
-```
-
-Learn a missing scenario. This is an AI workflow, not a single CLI command. Write a candidate patch; do not overwrite `business-map.json` directly.
-For a complex scenario with many sub-links, keep one scenario and add `flows` under it instead of creating many top-level scenarios.
-
-```bash
-/ckgraphify update --path <workspace-or-repos-root> [--graph <merged-graph.json>] [--map <business-map.json>]
-```
-
-Refresh graph-main outputs, rebuild the merged graph, validate the business map, and report stale anchors or missing repos.
-
 ```bash
 /ckgraphify help
 ```
@@ -60,26 +43,17 @@ Show this API briefly.
 
 - `--path` can point to a workspace, repos root, repo list, or a merged graph JSON. Resolve it before running CLI commands.
 - Two root directories:
-  - `${CLAUDE_PLUGIN_ROOT}` — installed plugin directory (graph data, map, evidence).
-  - `${CLAUDE_PROJECT_DIR}` — the project that installed the plugin (repos, pypi source, docs).
-- Graph and business map (read/write):
+  - `${CLAUDE_PLUGIN_ROOT}` — installed plugin directory (graph data, business map, and optional evidence).
+  - `${CLAUDE_PROJECT_DIR}` — optional user project directory for code reading only.
+- Graph and business map (read-only in the plugin):
   - Default business map is `${CLAUDE_PLUGIN_ROOT}/graphify-out/business-map.json`.
   - Default evidence shadow table is `${CLAUDE_PLUGIN_ROOT}/graphify-out/business-map-evidence.json`.
-  - Default merged graph is `${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json`.
-- Repos (code reading, main-graph generation):
+  - Default merged graph is `${CLAUDE_PLUGIN_ROOT}/graphify-out/graph.json`.
+- Repos (optional code reading only):
   - Default repos root is `${CLAUDE_PROJECT_DIR}/repos`.
-  - Each repo's graph-main output is `${CLAUDE_PROJECT_DIR}/repos/<repo>/graphify-out/graph-main.json`.
-- Docs and policies (read-only, bundled with pypi source):
-  - `${CLAUDE_PROJECT_DIR}/ckgraphify/docs/business-map-policy.md`
-- Sync: after writing to `${CLAUDE_PLUGIN_ROOT}/graphify-out/`, sync changes back to the project source:
-
-  ```bash
-  cp "${CLAUDE_PLUGIN_ROOT}/graphify-out/business-map.json" "${CLAUDE_PROJECT_DIR}/claude-plugin/graphify-out/business-map.json"
-  cp "${CLAUDE_PLUGIN_ROOT}/graphify-out/business-map-evidence.json" "${CLAUDE_PROJECT_DIR}/claude-plugin/graphify-out/business-map-evidence.json"
-  ```
+  - If `${CLAUDE_PROJECT_DIR}` is missing or repos are unavailable, answer from graph and business-map evidence and report that source-file reading is unavailable.
 
 - If `${CLAUDE_PLUGIN_ROOT}` is missing, report that the Claude plugin environment is not active.
-- If `${CLAUDE_PROJECT_DIR}` is missing, report that repos and docs are not available for code reading.
 - Default depth: `10`.
 - Default answer: concise business result first, evidence second, gaps last.
 - If required inputs are missing and cannot be inferred from files, ask one short clarification.
@@ -91,28 +65,25 @@ Prefer existing deterministic commands:
 ```bash
 "${GRAPHIFY:-graphify}" business-show --map <map> --concept <concept>
 "${GRAPHIFY:-graphify}" business-query "<question>" --map <map> --graph <graph> --trace --format text
-"${GRAPHIFY:-graphify}" business-lint --map <map>
-"${GRAPHIFY:-graphify}" business-validate --map <map> --graph <graph>
 "${GRAPHIFY:-graphify}" business-trace --map <map> --graph <graph> --concept <concept> --scenario <scenario> --flow <flow> --max-depth <n> --sources
 "${GRAPHIFY:-graphify}" main-trace --graph <graph> --from <exact-node> --api <api> --max-depth <n> --sources --prefer-repo <repo> --exclude-repo <repo>
-"${GRAPHIFY:-graphify}" main-graph "${CLAUDE_PROJECT_DIR}/repos/<repo>" --out "${CLAUDE_PROJECT_DIR}/repos/<repo>/graphify-out/graph-main.json"
-"${GRAPHIFY:-graphify}" merge-main-graphs "${CLAUDE_PROJECT_DIR}/repos/<repo-a>/graphify-out/graph-main.json" "${CLAUDE_PROJECT_DIR}/repos/<repo-b>/graphify-out/graph-main.json" --out "${CLAUDE_PLUGIN_ROOT}/graphify-out/graph-hc.json"
 ```
 
 ## Graph Scope
 
 Treat the merged graph as business navigation, not as a complete execution trace. Its normal responsibility is to get from business wording to the right repo, page, MTop/REST API, HSF method, dependency boundary, and source filepath.
 
-Do not expect `graph-hc.json` or other merged graphs to describe every parameter, response field, extension point, or method-local branch. That level of detail would make the graph too large and noisy. When a question depends on field-level or parameter-level behavior, first use `main-trace` and source filepaths to land on the right code, then read the relevant implementation directly.
+Do not expect `graph.json` or other merged graphs to describe every parameter, response field, extension point, or method-local branch. That level of detail would make the graph too large and noisy. When a question depends on field-level or parameter-level behavior, first use `main-trace` and source filepaths to land on the right code, then read the relevant implementation directly.
 
 ### File path convention
 
-Merged graphs store source paths as **relative paths** per repo. To resolve the full path for code reading:
+Merged graphs store source paths as **relative paths** per repo. To resolve the full path for optional code reading:
 
 - `graph.repo_roots` maps each repo name to its relative root, e.g. `{"p": "repos/p", "ele-newretail-drug": "repos/ele-newretail-drug"}`.
 - Each node has `repo` (repo name) and `source_file` (relative to the repo root).
 - Full path = `${CLAUDE_PROJECT_DIR}` + `/` + `repo_roots[node.repo]` + `/` + `node.source_file`.
 - `main-trace --sources` already outputs resolved paths in `repos/<repo>/<file>` format.
+- If `${CLAUDE_PROJECT_DIR}` or the target repo is not available, do not fail the query; cite the graph source path and report source reading as unavailable.
 
 Missing field-level graph evidence is not automatically a graph defect. Classify it as one of:
 
@@ -138,20 +109,6 @@ Search discipline:
 - If code reading is genuinely necessary, inspect only trace source files or a very small named set of files. Exclude generated graph output, HTML, caches, and unrelated repo directories.
 - Prefer additional graphify tool calls over raw shell searches when adjusting scope, parameters, or neighboring scenarios.
 
-For `/ckgraphify learn`:
-
-1. Require concept, scenario, repos, graph, map, and out.
-2. Read `${CLAUDE_PROJECT_DIR}/ckgraphify/docs/business-map-policy.md`; every map iteration must follow it or explicitly report the conflict.
-3. Read existing map scenarios, flows, gaps, accepted gaps, verification scopes, and `search_constraints`.
-4. Search graph nodes with aliases, keywords, page/API names, HSF names, and repo hints.
-5. Run `main-trace` on promising graph matches.
-6. Validate important links in code, especially when the answer depends on response fields, request parameters, extension points, or method-local branching.
-7. For a simple branch, write a candidate patch to `--out` with scenario-level anchors, trace hints, verification scope, boundaries, and remaining gaps. Keep detailed source evidence in `${CLAUDE_PLUGIN_ROOT}/graphify-out/business-map-evidence.json`, not in `business-map.json`.
-8. For a complex branch, write one scenario with `flows`; each flow may have `id`, `name`, `status`, `summary`, `verification_scope`, `anchors`, `trace_hints`, `gaps`, and `accepted_gaps`. Keep `business-map.json` compact; use the evidence shadow table for source proof.
-9. Do not split a user's named complex scenario into multiple top-level scenarios unless the user asks for that model.
-10. Run `business-lint` and `business-validate`; run `business-trace` if hints are usable.
-11. After writing to `${CLAUDE_PLUGIN_ROOT}/graphify-out/`, sync changes back to `${CLAUDE_PROJECT_DIR}/claude-plugin/graphify-out/`.
-
 ## Answer Contract
 
 Include:
@@ -168,16 +125,8 @@ Keep graph-main schema details, dependency extraction rules, and low-level branc
 
 ## Guardrails
 
-- Do not write business semantics into `graph-main.json`.
-- Before changing `business-map.json`, read `${CLAUDE_PROJECT_DIR}/ckgraphify/docs/business-map-policy.md`; follow it or state the conflict.
-- Use `scenario.flows[]` for sub-links inside one complex business scene, such as landing page purchase/refund/benefit branches.
-- Keep `business-map.json` as a search/trace index. Do not add long `evidence` arrays there; put audit proof in `graphify-out/business-map-evidence.json`.
-- Use `gaps` for active unknowns and `accepted_gaps` for known boundaries with a stop rule.
-- New learned scenarios go to candidate patch files first.
+- Treat the Claude plugin as a read-only consumer of graph and business-map artifacts.
+- Do not edit `graph-main.json`, `graph.json`, `business-map.json`, or `business-map-evidence.json` from this skill.
+- Do not run learn, update, lint, validate, main-graph, or merge-main-graphs as part of answering user queries.
 - Missing graph evidence is a coverage gap, not proof that code behavior does not exist.
-- After modifying only `business-map.json`, run `graphify business-lint --map <map>` and `graphify business-validate --map <map> --graph <graph>`; `graphify update .` is not required for a map-only semantic edit.
-- After modifying code, docs, skills, extraction, merge, trace, business-map tooling, or HTML output, run:
-
-```bash
-graphify update .
-```
+- If the answer needs knowledge not covered by the current artifacts, report the gap and suggest that the server-side kg production/learning workflow refresh the graph or business map.