contextlake 2.2.0__tar.gz → 2.3.0__tar.gz

{contextlake-2.2.0/src/contextlake.egg-info → contextlake-2.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: contextlake
-Version: 2.2.0
+Version: 2.3.0
 Summary: A local context layer for AI tools: mirror your repositories, index them into a knowledge graph, and serve it over MCP.
 Author-email: Sayak Sarkar <sayak.bugsmith@gmail.com>
 License-Expression: MIT
@@ -44,13 +44,20 @@ Requires-Dist: llama-cpp-python>=0.2; extra == "llm-local"
 Requires-Dist: huggingface_hub>=0.20; extra == "llm-local"
 Dynamic: license-file
 
-# contextlake
+<p align="center">
+  <img src="docs/branding/glyph.svg" alt="" width="76" height="76">
+</p>
+<h1 align="center">contextlake</h1>
+<p align="center"><em>All your real context, in one local lake.</em></p>
 
-> **A local context layer for your AI tools — your repositories mirrored, indexed into a knowledge graph, and served over MCP, so agents work from real source instead of guessing.**
+<p align="center">
+  <a href="https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml"><img src="https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
+  <a href="https://pypi.org/project/contextlake/"><img src="https://img.shields.io/pypi/v/contextlake?color=137A8B" alt="PyPI"></a>
+  <img src="https://img.shields.io/badge/python-3.9%2B-blue" alt="Python 3.9+">
+  <img src="https://img.shields.io/badge/license-MIT-green" alt="License: MIT">
+</p>
 
-[![CI](https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml/badge.svg)](https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml)
-![Python](https://img.shields.io/badge/python-3.9%2B-blue)
-![License: MIT](https://img.shields.io/badge/license-MIT-green)
+> **A local context layer for your AI tools — your repositories mirrored, indexed into a knowledge graph, and served over MCP, so agents work from real source instead of guessing.**
 
 You have access to dozens — maybe hundreds — of repositories scattered across a
 GitLab group and its subgroups. You want them all on your laptop, in the same
@@ -135,6 +142,11 @@ The tool carries no credentials of its own — auth rides on `glab` — so
 `.contextlake.ini` holds only non-secret settings and is gitignored by default. The
 full option reference is in [docs/usage.md](docs/usage.md).
 
+> **Behind a slow / TLS-inspecting corporate proxy** (e.g. Zscaler) where `glab`'s API
+> calls time out, set `GITLAB_TOKEN` (a `read_api` token) — contextlake then enumerates
+> projects via its own HTTP client, which tolerates the slow DNS where `glab`'s short
+> dial timeout fails.
+
 ## Usage
 
 Run commands as `contextlake <command>` — full per-command docs are in
@@ -150,7 +162,8 @@ Run commands as `contextlake <command>` — full per-command docs are in
 | `update` | Pull updates for local repos (skips only repos with a dirty working tree) |
 | `branches` | Switch each repo to its most active branch |
 | `verify` | Check the local mirror matches GitLab (drift, orphans, nesting) |
-| `sync` | The full pipeline: fetch → clone → update → branches → verify |
+| `sync` | The full pipeline: fetch → clone → update → branches → verify → audit |
+| `audit` | Repo health & age: empty/README-only repos + creation & last-commit dates (JSON + CSV) |
 | `bootstrap` | **Turnkey**: sync + index + connect + embed + wiki + steer |
 | `index` | Build the code/dependency graph (`--workspace`, incremental, `--watch`) |
 | `connect` | Link repos to Atlassian / Figma / GitLab sources |
@@ -160,9 +173,10 @@ Run commands as `contextlake <command>` — full per-command docs are in
 | `steer` | Write editor steering — `AGENTS.md`, `.mcp.json`, `.windsurfrules`, skills |
 | `serve` | Expose the graph over MCP (`--transport stdio`/`http`) |
 | `query` | Search the index (`--kind`, `--repo`, `--limit`, `--as-of <commit>`) |
+| `graph` | Visualize the graph — offline interactive HTML / DOT / Mermaid / JSON (`--overview`, `--serve`) |
 | `doctor` | Check the knowledge-layer environment (SQLite FTS5, git/glab, store, embeddings) |
 
-The first seven are the core sync (detailed below); the rest are the optional
+The first eight are the core sync (detailed below); the rest are the optional
 **[knowledge layer](#knowledge-layer)**. Run any command with `--config` (sync INI)
 and, for the knowledge layer, `--config`/`--kb-config` pointing at your `kb.toml`.
 
@@ -194,8 +208,10 @@ Beyond mirroring, an optional layer (`contextlake.kb`) turns your repos into a
 **knowledge graph** and serves it to AI tools over **MCP** — so Claude Code, Windsurf,
 or Kiro can answer *"where is `X` defined?"* or *"who calls `Y`?"* instead of grepping.
 It can also link repos to their Atlassian / Figma / GitLab items, add semantic search,
-write a curated wiki, and generate per-tool steering files + a skills library. Most of
-it needs no model; the rest works with a local Ollama or any OpenAI-compatible endpoint.
+write a curated wiki, **visualize the graph** (`contextlake graph` → an offline, interactive
+HTML — fleet overview, a symbol's neighbourhood, or a single repo), and generate per-tool
+steering files + a skills library. Most of it needs no model; the rest works with a local
+Ollama or any OpenAI-compatible endpoint.
 
 One command sets it all up:
 

{contextlake-2.2.0 → contextlake-2.3.0}/README.md

@@ -1,11 +1,18 @@
-# contextlake
+<p align="center">
+  <img src="docs/branding/glyph.svg" alt="" width="76" height="76">
+</p>
+<h1 align="center">contextlake</h1>
+<p align="center"><em>All your real context, in one local lake.</em></p>
+
+<p align="center">
+  <a href="https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml"><img src="https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml/badge.svg" alt="CI"></a>
+  <a href="https://pypi.org/project/contextlake/"><img src="https://img.shields.io/pypi/v/contextlake?color=137A8B" alt="PyPI"></a>
+  <img src="https://img.shields.io/badge/python-3.9%2B-blue" alt="Python 3.9+">
+  <img src="https://img.shields.io/badge/license-MIT-green" alt="License: MIT">
+</p>
 
 > **A local context layer for your AI tools — your repositories mirrored, indexed into a knowledge graph, and served over MCP, so agents work from real source instead of guessing.**
 
-[![CI](https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml/badge.svg)](https://github.com/sayak-sarkar/contextlake/actions/workflows/ci.yml)
-![Python](https://img.shields.io/badge/python-3.9%2B-blue)
-![License: MIT](https://img.shields.io/badge/license-MIT-green)
-
 You have access to dozens — maybe hundreds — of repositories scattered across a
 GitLab group and its subgroups. You want them all on your laptop, in the same
 shape they have on GitLab, each sitting on the branch where the real work is
@@ -89,6 +96,11 @@ The tool carries no credentials of its own — auth rides on `glab` — so
 `.contextlake.ini` holds only non-secret settings and is gitignored by default. The
 full option reference is in [docs/usage.md](docs/usage.md).
 
+> **Behind a slow / TLS-inspecting corporate proxy** (e.g. Zscaler) where `glab`'s API
+> calls time out, set `GITLAB_TOKEN` (a `read_api` token) — contextlake then enumerates
+> projects via its own HTTP client, which tolerates the slow DNS where `glab`'s short
+> dial timeout fails.
+
 ## Usage
 
 Run commands as `contextlake <command>` — full per-command docs are in
@@ -104,7 +116,8 @@ Run commands as `contextlake <command>` — full per-command docs are in
 | `update` | Pull updates for local repos (skips only repos with a dirty working tree) |
 | `branches` | Switch each repo to its most active branch |
 | `verify` | Check the local mirror matches GitLab (drift, orphans, nesting) |
-| `sync` | The full pipeline: fetch → clone → update → branches → verify |
+| `sync` | The full pipeline: fetch → clone → update → branches → verify → audit |
+| `audit` | Repo health & age: empty/README-only repos + creation & last-commit dates (JSON + CSV) |
 | `bootstrap` | **Turnkey**: sync + index + connect + embed + wiki + steer |
 | `index` | Build the code/dependency graph (`--workspace`, incremental, `--watch`) |
 | `connect` | Link repos to Atlassian / Figma / GitLab sources |
@@ -114,9 +127,10 @@ Run commands as `contextlake <command>` — full per-command docs are in
 | `steer` | Write editor steering — `AGENTS.md`, `.mcp.json`, `.windsurfrules`, skills |
 | `serve` | Expose the graph over MCP (`--transport stdio`/`http`) |
 | `query` | Search the index (`--kind`, `--repo`, `--limit`, `--as-of <commit>`) |
+| `graph` | Visualize the graph — offline interactive HTML / DOT / Mermaid / JSON (`--overview`, `--serve`) |
 | `doctor` | Check the knowledge-layer environment (SQLite FTS5, git/glab, store, embeddings) |
 
-The first seven are the core sync (detailed below); the rest are the optional
+The first eight are the core sync (detailed below); the rest are the optional
 **[knowledge layer](#knowledge-layer)**. Run any command with `--config` (sync INI)
 and, for the knowledge layer, `--config`/`--kb-config` pointing at your `kb.toml`.
 
@@ -148,8 +162,10 @@ Beyond mirroring, an optional layer (`contextlake.kb`) turns your repos into a
 **knowledge graph** and serves it to AI tools over **MCP** — so Claude Code, Windsurf,
 or Kiro can answer *"where is `X` defined?"* or *"who calls `Y`?"* instead of grepping.
 It can also link repos to their Atlassian / Figma / GitLab items, add semantic search,
-write a curated wiki, and generate per-tool steering files + a skills library. Most of
-it needs no model; the rest works with a local Ollama or any OpenAI-compatible endpoint.
+write a curated wiki, **visualize the graph** (`contextlake graph` → an offline, interactive
+HTML — fleet overview, a symbol's neighbourhood, or a single repo), and generate per-tool
+steering files + a skills library. Most of it needs no model; the rest works with a local
+Ollama or any OpenAI-compatible endpoint.
 
 One command sets it all up:
 

{contextlake-2.2.0 → contextlake-2.3.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "contextlake"
-version = "2.2.0"
+version = "2.3.0"
 description = "A local context layer for AI tools: mirror your repositories, index them into a knowledge graph, and serve it over MCP."
 readme = "README.md"
 license = "MIT"
@@ -72,9 +72,9 @@ package-dir = { "" = "src" }
 where = ["src"]
 
 [tool.setuptools.package-data]
-# Vendored, MIT-licensed cytoscape.js for the offline `contextlake graph` HTML.
-# Without this the wheel silently omits the asset (works from a checkout only).
-"contextlake.kb" = ["static/*.js"]
+# Vendored cytoscape.js + the extracted app.css/app.js for the offline
+# `contextlake graph` HTML. Without this the wheel silently omits them.
+"contextlake.kb" = ["static/*.js", "static/*.css"]
 
 [tool.pytest.ini_options]
 testpaths = ["tests"]

{contextlake-2.2.0 → contextlake-2.3.0}/src/contextlake/__init__.py

@@ -1,3 +1,3 @@
 """contextlake — keep a local workspace mirrored with GitLab repositories."""
 
-__version__ = "2.2.0"
+__version__ = "2.3.0"

contextlake-2.3.0/src/contextlake/kb/arch/__init__.py

@@ -0,0 +1 @@
+"""Architecture-level resolution & diagram export over the code graph."""

contextlake-2.3.0/src/contextlake/kb/arch/resolve.py

@@ -0,0 +1,44 @@
+"""Repo-level architecture resolution from the code knowledge graph.
+
+The **trustworthy** cross-repo signal is the *package two-hop*: repo A
+**publishes** a package that repo B **depends_on** → B depends on A. Raw
+cross-repo ``imports`` edges are dominated by import-star artifacts (global
+``module`` nodes like ``System``/``xunit`` shared across the fleet), so they are
+deliberately NOT used here. The result is **inferred** (a manifest-derived,
+likely-undercount signal), never presented as ground truth.
+
+Stdlib-only; one SQL query against the shared store.
+"""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from ..store.base import Store
+
+# dependent_repo --depends_on--> publisher_repo, weighted by shared package count.
+_TWO_HOP = """
+SELECT dep.dep_repo, pub.pub_repo, COUNT(DISTINCT pub.pkg) AS shared
+FROM (SELECT np.repo_id AS pub_repo, e.dst AS pkg FROM edges e
+        JOIN nodes np ON np.node_id = e.src WHERE e.relation = 'publishes') pub
+JOIN (SELECT nd.repo_id AS dep_repo, e.dst AS pkg FROM edges e
+        JOIN nodes nd ON nd.node_id = e.src WHERE e.relation = 'depends_on') dep
+  ON pub.pkg = dep.pkg
+WHERE pub.pub_repo != dep.dep_repo
+GROUP BY dep.dep_repo, pub.pub_repo
+"""
+
+
+def repo_dependency_edges(store: Store) -> list[dict]:
+    """Real repo→repo dependencies via the package two-hop (``publishes ⨝ depends_on``).
+
+    Each edge is ``dependent --depends_on--> publisher``, ``weight`` = number of
+    shared packages, marked ``INFERRED`` (manifest-derived, a likely undercount —
+    not every dependency declares/publishes a package). Far smaller and far more
+    trustworthy than the raw cross-repo ``imports`` edges.
+    """
+    rows = store.conn.execute(_TWO_HOP).fetchall()
+    return [{"src": dep, "dst": pub, "relation": "depends_on",
+             "confidence": "INFERRED", "weight": shared}
+            for dep, pub, shared in rows]

{contextlake-2.2.0 → contextlake-2.3.0}/src/contextlake/kb/commands.py

@@ -840,12 +840,13 @@ def cmd_graph(args) -> int:
         # is findable); neighbourhood/repo views stay bounded at 500.
         max_nodes = getattr(args, "max_nodes", None) or (5000 if overview else 500)
 
+        meta: dict = {}
         if overview:
-            nodes, edges = viz.overview_subgraph(store, max_nodes=max_nodes)
-            meta = {"mode": "overview"}
+            nodes, edges = viz.overview_subgraph(store, max_nodes=max_nodes, meta=meta)
+            meta["mode"] = "overview"
         elif getattr(args, "repo", None) and not _has_seed(args):
-            nodes, edges = viz.repo_subgraph(store, args.repo, max_nodes=max_nodes)
-            meta = {"mode": "repo", "repo": args.repo}
+            nodes, edges = viz.repo_subgraph(store, args.repo, max_nodes=max_nodes, meta=meta)
+            meta.update(mode="repo", repo=args.repo)
         else:
             seeds = viz.seed_ids_from_args(store, args)
             if not seeds:
@@ -855,8 +856,8 @@ def cmd_graph(args) -> int:
             nodes, edges = viz.extract_subgraph(
                 store, seeds, hops=hops, max_nodes=max_nodes, max_fanout=max_fanout,
                 relation=getattr(args, "relation", None),
-                direction=getattr(args, "direction", None) or "both")
-            meta = {"mode": "neighborhood", "seed_ids": seeds, "hops": hops}
+                direction=getattr(args, "direction", None) or "both", meta=meta)
+            meta.update(mode="neighborhood", seed_ids=seeds, hops=hops)
 
         payload = viz.to_payload(nodes, edges, meta)
         cdn = getattr(args, "cdn", False)

contextlake-2.3.0/src/contextlake/kb/static/app.css

@@ -0,0 +1,137 @@
+
+  :root{
+    --deepwater:#0E2A33;--lake:#137A8B;--current:#2BB3A3;--mist:#EAF4F4;
+    --shore:#D7C5A0;--sun:#E7B53C;
+    --bg:#f5fafb;--surface:#ffffff;--surface-2:#eef6f7;--line:#dce8ea;--line-2:#c6d8db;
+    --text:#0E2A33;--muted:#5b7177;--subtle:#8aa2a6;--brand:#137A8B;--brand-h:#0f6473;
+    --accent:#2BB3A3;--focus:#2BB3A3;--canvas-label:#0E2A33;--surface-solid:#ffffff;
+    --faded-opacity:0.08;
+    --cy-bg:radial-gradient(120% 90% at 50% -10%,#ffffff 0%,#f1fafb 45%,#e3f1f2 100%);
+    --e1:0 1px 2px rgba(14,42,51,.07);--e2:0 4px 14px -4px rgba(14,42,51,.18);
+    --rail:288px;--inspect:0px;--bar:48px;--status:30px;
+    --ff:"Inter",system-ui,-apple-system,Segoe UI,Roboto,sans-serif;
+    --ff-d:"Space Grotesk",var(--ff);--dur:.16s;--ease:cubic-bezier(.2,.6,.2,1);
+  }
+  body[data-theme="dark"]{
+    --bg:#0a1f26;--surface:#102e38;--surface-2:#16414e;--line:#1d4d5b;--line-2:#286675;
+    --text:#EAF4F4;--muted:#9bbcc2;--subtle:#6f969e;--brand:#2BB3A3;--brand-h:#3fc6b6;
+    --canvas-label:#EAF4F4;--surface-solid:#102e38;
+    --faded-opacity:0.16;
+    --cy-bg:radial-gradient(120% 90% at 50% -10%,#103039 0%,#0c252d 45%,#081b21 100%);
+    --e1:0 1px 2px rgba(0,0,0,.4);--e2:0 6px 18px -6px rgba(0,0,0,.55);
+  }
+  body[data-sidebar="collapsed"]{--rail:0px}
+  body[data-inspect="open"]{--inspect:340px}
+  *{box-sizing:border-box}
+  html,body{margin:0;height:100%;color:var(--text);background:var(--bg);font-family:var(--ff)}
+  #app{display:grid;height:100%;grid-template-columns:var(--rail) 1fr var(--inspect);
+    grid-template-rows:var(--bar) 1fr var(--status);
+    grid-template-areas:"top top top" "side cy inspect" "status status status";
+    transition:grid-template-columns var(--dur) var(--ease)}
+  #topbar{grid-area:top;display:flex;align-items:center;gap:10px;padding:0 12px;
+    background:var(--surface);border-bottom:1px solid var(--line);z-index:5}
+  .glyph{width:26px;height:26px;border-radius:7px;display:block;box-shadow:var(--e1)}
+  .wm{font-family:var(--ff-d);font-size:16px;font-weight:600;letter-spacing:-.01em;color:var(--text)}
+  .wm .l{color:var(--brand)}
+  #mode{font-size:11px;color:var(--muted);padding:2px 9px;border:1px solid var(--line);
+    border-radius:999px;background:var(--surface-2);text-transform:capitalize}
+  .grow{flex:1}
+  .tsearch{position:relative;display:flex;align-items:center;width:260px;max-width:34vw}
+  .tsearch .si{position:absolute;left:9px;width:14px;height:14px;color:var(--subtle)}
+  #search{width:100%;padding:7px 10px 7px 28px;font-size:12px;border:1px solid var(--line);
+    border-radius:8px;background:var(--surface-2);color:var(--text);
+    transition:border-color var(--dur),box-shadow var(--dur)}
+  #search:focus{outline:none;border-color:var(--brand);box-shadow:0 0 0 3px rgba(43,179,163,.25);
+    background:var(--surface)}
+  #panel{grid-area:side;overflow-y:auto;background:var(--surface);
+    border-right:1px solid var(--line);padding:14px;font-size:12px;min-width:0}
+  body[data-sidebar="collapsed"] #panel{padding:0;border:0;overflow:hidden}
+  .sgroup{margin-bottom:16px}
+  .sgroup h2{font-size:10px;font-weight:600;text-transform:uppercase;letter-spacing:.06em;
+    color:var(--muted);margin:0 0 8px}
+  .row{display:flex;gap:6px;align-items:center;flex-wrap:wrap}
+  #viewmodes{margin-bottom:9px}
+  #viewmodes[hidden]{display:none}
+  .seg{display:inline-flex;border:1px solid var(--line);border-radius:8px;overflow:hidden;
+    background:var(--surface-2)}
+  .segbtn{font:inherit;font-size:12px;padding:6px 13px;border:0;background:transparent;
+    color:var(--muted);cursor:pointer;transition:background var(--dur),color var(--dur)}
+  .segbtn+.segbtn{border-left:1px solid var(--line)}
+  .segbtn.on{background:var(--brand);color:#fff;font-weight:500}
+  .segbtn:not(.on):hover{background:var(--surface);color:var(--text)}
+  .tog{display:flex;align-items:center;gap:6px;margin-top:9px;font-size:11px;
+    color:var(--muted);cursor:pointer}
+  .tog[hidden]{display:none}
+  .tog input{accent-color:var(--brand);cursor:pointer;margin:0}
+  .tog .cnt{color:var(--subtle);font-variant-numeric:tabular-nums}
+  label{font-size:11px;color:var(--muted)}
+  select,.btn,.ibtn{font:inherit;font-size:12px;border:1px solid var(--line);border-radius:8px;
+    background:var(--surface);color:var(--text);cursor:pointer;
+    transition:background var(--dur),border-color var(--dur),box-shadow var(--dur)}
+  select{padding:6px 8px}
+  .btn{padding:6px 11px;display:inline-flex;align-items:center;gap:6px}
+  .ibtn{padding:6px;width:30px;height:30px;display:inline-flex;align-items:center;
+    justify-content:center}
+  .ibtn svg,.btn svg{width:15px;height:15px}
+  .btn:hover,.ibtn:hover{background:var(--surface-2);border-color:var(--line-2)}
+  .btn:active,.ibtn:active{transform:translateY(1px)}
+  .btn.primary{background:var(--brand);border-color:var(--brand);color:#fff;box-shadow:var(--e1)}
+  .btn.primary:hover{background:var(--brand-h);border-color:var(--brand-h)}
+  :focus-visible{outline:none;box-shadow:0 0 0 2px var(--surface),0 0 0 4px var(--focus)}
+  #legend,#edgelegend{display:flex;flex-wrap:wrap;gap:5px}
+  .lg{display:inline-flex;align-items:center;gap:5px;padding:3px 8px;border-radius:999px;
+    border:1px solid var(--line);background:var(--surface);color:var(--text);cursor:pointer;
+    user-select:none;font-size:11px;transition:background var(--dur),border-color var(--dur)}
+  .lg:hover{background:var(--surface-2);border-color:var(--line-2)}
+  .lg i{width:9px;height:9px;border-radius:50%;display:inline-block}
+  .lg.rel i{border-radius:1px;width:12px;height:3px}
+  .lg .cnt{color:var(--subtle);font-variant-numeric:tabular-nums;font-size:10px}
+  .lg.off{opacity:.4}
+  .lg.off .lbl{text-decoration:line-through}
+  #cy{grid-area:cy;min-width:0;min-height:0;position:relative;background:var(--cy-bg)}
+  #info{grid-area:inspect;overflow-y:auto;background:var(--surface);
+    border-left:1px solid var(--line);padding:14px;font-size:12px;display:none}
+  body[data-inspect="open"] #info{display:block}
+  #info h2{font-size:14px;margin:0 0 2px;color:var(--text);word-break:break-all}
+  #info dl{display:grid;grid-template-columns:auto 1fr;gap:4px 10px;margin:9px 0 0;
+    border-top:1px solid var(--line);padding-top:9px}
+  #info dt{color:var(--brand);font-weight:500;font-size:11px}
+  #info dd{margin:0;word-break:break-all;font-size:11px}
+  #info .hint{color:var(--subtle);margin-top:9px;font-style:italic}
+  #info .edge-flow{font-size:11px;color:var(--muted);margin:5px 0 2px;word-break:break-all}
+  #info .trust{display:flex;align-items:center;gap:6px;margin:8px 0 2px;font-size:11px}
+  #info .trust .dot{width:9px;height:9px;border-radius:50%;display:inline-block}
+  #info .trust .blurb{color:var(--muted)}
+  #info .copy-prov{margin-top:10px}
+  #info .conns{margin-top:11px;border-top:1px solid var(--line);padding-top:9px}
+  #info .conns h3{font-size:10px;font-weight:600;text-transform:uppercase;letter-spacing:.06em;
+    color:var(--muted);margin:0 0 7px}
+  #info .conns .cc{color:var(--subtle);font-weight:500}
+  #info .conns ul{list-style:none;margin:0;padding:0;display:flex;flex-direction:column;gap:5px}
+  #info .conns li{display:flex;align-items:center;gap:6px;font-size:11px;line-height:1.3}
+  #info .conns .rdot{width:8px;height:8px;border-radius:50%;flex:none}
+  #info .conns .rl{color:var(--muted);white-space:nowrap;flex:none}
+  #info .conns .rn{color:var(--brand);cursor:pointer;word-break:break-all;
+    text-decoration:underline;text-decoration-color:transparent;transition:text-decoration-color var(--dur)}
+  #info .conns .rn:hover{text-decoration-color:currentColor}
+  #info .conns .rmore{color:var(--brand);font-style:italic;cursor:pointer}
+  #info .conns .rmore:hover{text-decoration:underline}
+  .rel-chip{display:inline-block;padding:2px 9px;border-radius:999px;color:#fff;
+    font-size:12px;font-weight:600}
+  #statusbar{grid-area:status;display:flex;align-items:center;gap:10px;padding:0 12px;
+    background:var(--surface);border-top:1px solid var(--line);font-size:11px;color:var(--muted)}
+  #statusbar .l{color:var(--brand)}
+  .trunc{color:var(--sun);font-weight:600;display:none}
+  .trunc.show{display:inline}
+  #tip{position:absolute;z-index:40;pointer-events:none;display:none;background:var(--deepwater);
+    color:var(--mist);font-size:11px;padding:4px 8px;border-radius:6px;max-width:42ch;
+    box-shadow:var(--e2)}
+  #empty{position:absolute;inset:0;display:none;flex-direction:column;align-items:center;
+    justify-content:center;gap:6px;color:var(--muted);text-align:center;padding:20px}
+  #empty.show{display:flex}
+  #empty .et{font-size:14px;color:var(--text);font-weight:600}
+  #empty code{background:var(--surface-2);padding:1px 5px;border-radius:4px;font-size:11px}
+  #cy.loading::after{content:"";position:absolute;top:0;left:0;height:2px;width:28%;
+    background:var(--accent);animation:ld 1s var(--ease) infinite}
+  @keyframes ld{0%{left:-28%}100%{left:100%}}
+  @media (prefers-reduced-motion:reduce){*{transition:none!important;animation:none!important}}