docs(readme): polish launch highlights and banner (#261)

* docs: simplify README launch highlights

* docs(readme): use six launch highlights

* docs(readme): use optimized banner asset

* ci: lint pull request titles

.github/workflows/commits.yml

@@ -13,6 +13,18 @@ concurrency:
   cancel-in-progress: true
 
 jobs:
+  title:
+    name: pull request title
+    if: github.event_name == 'pull_request'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Validate Conventional Commit PR title
+        env:
+          PR_TITLE: ${{ github.event.pull_request.title }}
+        run: make check-pr-title
+
   messages:
     name: commit messages
     runs-on: ubuntu-latest

Makefile

@@ -1,4 +1,4 @@
-.PHONY: help install install-deps lint docs-check check-commits check-assets check-deprecated-names check-cjk check-datetime openapi check-openapi format test integration package cov ci clean
+.PHONY: help install install-deps lint docs-check check-commits check-pr-title check-assets check-deprecated-names check-cjk check-datetime openapi check-openapi format test integration package cov ci clean
 
 help:
 	@echo "Targets:"
@@ -7,6 +7,7 @@ help:
 	@echo "  lint          ruff + import-linter + repo hygiene + datetime discipline + openapi drift"
 	@echo "  docs-check    Validate Markdown links, use-case banners, and issue template YAML"
 	@echo "  check-commits Validate Conventional Commit subjects for a git range"
+	@echo "  check-pr-title Validate PR title uses Conventional Commit format"
 	@echo "  check-assets  Block committed images, videos, and asset/media directories"
 	@echo "  check-deprecated-names Block deprecated product names"
 	@echo "  check-cjk     Scan for CJK outside the language-policy allowlist (advisory)"
@@ -48,6 +49,9 @@ docs-check:
 check-commits:
 	python3 scripts/check_commit_messages.py $(RANGE)
 
+check-pr-title:
+	python3 scripts/check_pr_title.py
+
 # Repository media hygiene gate. Images/videos belong in external hosting,
 # release artifacts, or other approved storage, then linked from docs.
 check-assets:

README.md

@@ -1,6 +1,6 @@
 <div align="center" id="readme-top">
 
-![EverOS banner](https://github.com/EverMind-AI/EverOS/releases/download/v1.0.0/everos-readme-banner.jpg)
+![EverOS banner](https://github.com/EverMind-AI/EverOS/releases/download/v1.0.0/everos-readme-banner-optimized.jpg)
 
 <p align="center">
   <a href="https://x.com/evermind"><img src="https://img.shields.io/badge/EverMind-000000?labelColor=gray&style=for-the-badge&logo=x&logoColor=white" alt="X"></a>
@@ -54,53 +54,36 @@
 <table>
 <tr>
 <td width="33%" valign="top">
-<strong>Markdown-First Memory</strong><br>
+<strong>Markdown As Source Of Truth</strong><br>
-Memory is persisted as plain Markdown: visible, auditable, hand-editable,
+All memory is persisted as <code>.md</code> files: readable, editable,
-Git-friendly, and owned by the user.
+grep-able, Git-versioned, and openable directly in Obsidian.
 </td>
 <td width="33%" valign="top">
-<strong>Lightweight Local Stack</strong><br>
+<strong>Local Three-Part Stack</strong><br>
-Install with Python. SQLite tracks runtime state; LanceDB powers vector,
+Markdown + SQLite + LanceDB keep vectors, BM25, and scalar filters
-BM25, and scalar-filter retrieval locally.
+local. No MongoDB, Elasticsearch, or Redis required.
 </td>
 <td width="33%" valign="top">
-<strong>Layered Memory Model</strong><br>
+<strong>Dual-Track Memory</strong><br>
-User memory and agent memory are first-class today. Wiki-style knowledge
+Agent memory (<code>cases</code> / <code>skills</code>) and user memory
-is the next layer in the roadmap.
+(<code>episodes</code> / <code>profile</code>) are extracted independently.
 </td>
 </tr>
 <tr>
 <td width="33%" valign="top">
-<strong>Self-Evolving Agents</strong><br>
-Agent memory can extract reusable cases and skills from repeated
-experience, so workflows become smarter over time.
-</td>
-<td width="33%" valign="top">
 <strong>Multimodal Ingestion</strong><br>
-Text, image, audio, documents, PDF, HTML, and email can be parsed into
+Text, images, audio, documents, PDFs, HTML, and email are unified into
-memory through the optional multimodal pipeline.
+searchable memory.
 </td>
 <td width="33%" valign="top">
-<strong>Online And Offline Strategy Control</strong><br>
+<strong>Self-Evolution</strong><br>
-Online extraction and offline evolution stay separate, with configurable
+Common skills are extracted from real usage; repeated patterns become
-prompts and models at each step. Dreaming is coming next.
+reusable workflows, no retraining required.
-</td>
-</tr>
-<tr>
-<td width="33%" valign="top">
-<strong>Orthogonal Memory Scope</strong><br>
-Owner, memory type, and scope are independent: search by user, agent,
-app, project, session, and structured filters.
 </td>
 <td width="33%" valign="top">
-<strong>Progressive Disclosure</strong><br>
+<strong>Orthogonal Retrieval</strong><br>
-Readable memory surfaces stay simple while deeper facts, cases, and
+Search independently by <code>user_id</code>, <code>agent_id</code>,
-skills remain available.
+<code>app_id</code>, <code>project_id</code>, and <code>session_id</code>.
-</td>
-<td width="33%" valign="top">
-<strong>Modular By Design</strong><br>
-EverAlgo owns algorithms; EverOS owns runtime, persistence, online flows,
-and offline evolution.
 </td>
 </tr>
 </table>

README.zh-CN.md

@@ -1,6 +1,6 @@
 <div align="center" id="readme-top">
 
-![EverOS banner](https://github.com/EverMind-AI/EverOS/releases/download/v1.0.0/everos-readme-banner.jpg)
+![EverOS banner](https://github.com/EverMind-AI/EverOS/releases/download/v1.0.0/everos-readme-banner-optimized.jpg)
 
 <p align="center">
   <a href="https://x.com/evermind"><img src="https://img.shields.io/badge/EverMind-000000?labelColor=gray&style=for-the-badge&logo=x&logoColor=white" alt="X"></a>
@@ -54,44 +54,30 @@
 <table>
 <tr>
 <td width="33%" valign="top">
-<strong>Markdown-First Memory</strong><br>
+<strong>Markdown As Source Of Truth</strong><br>
-记忆以普通 Markdown 持久化：可见、可审计、可手动编辑、Git 友好，并由用户自己拥有。
+所有记忆持久化为 <code>.md</code> 文件：可读、可改、可 grep、可 Git 版本化，也可直接用 Obsidian 打开。
 </td>
 <td width="33%" valign="top">
-<strong>Lightweight Local Stack</strong><br>
+<strong>Local Three-Part Stack</strong><br>
-用 Python 即可安装。SQLite 负责运行时状态；LanceDB 在本地提供向量、BM25 和结构化过滤检索。
+Markdown + SQLite + LanceDB 在本地完成向量、BM25 和标量过滤检索，无需 MongoDB、Elasticsearch 或 Redis。
 </td>
 <td width="33%" valign="top">
-<strong>Layered Memory Model</strong><br>
+<strong>Dual-Track Memory</strong><br>
-用户记忆和 Agent 记忆现在是一等公民。Wiki 式知识层是路线图中的下一层。
+Agent 记忆（<code>cases</code> / <code>skills</code>）与用户记忆（<code>episodes</code> / <code>profile</code>）独立提取，互不污染。
 </td>
 </tr>
 <tr>
 <td width="33%" valign="top">
-<strong>Self-Evolving Agents</strong><br>
-Agent 记忆可以从重复经验中提取可复用的 cases 和 skills，让工作流随着时间变得更聪明。
-</td>
-<td width="33%" valign="top">
 <strong>Multimodal Ingestion</strong><br>
-文本、图片、音频、文档、PDF、HTML 和邮件都可以通过可选的多模态管线解析进记忆。
+文本、图像、音频、文档、PDF、HTML 和邮件统一抽取为可检索的记忆形态。
 </td>
 <td width="33%" valign="top">
-<strong>Online And Offline Strategy Control</strong><br>
+<strong>Self-Evolution</strong><br>
-在线提取和离线进化保持分离，并且每一步都可以配置 prompts 和 models。Dreaming 即将到来。
+从真实使用经验中自动抽取共性 skills，重复模式沉淀为可复用流程，无需重训。
-</td>
-</tr>
-<tr>
-<td width="33%" valign="top">
-<strong>Orthogonal Memory Scope</strong><br>
-Owner、memory type 和 scope 相互独立：可以按 user、agent、app、project、session 和结构化 filters 搜索。
 </td>
 <td width="33%" valign="top">
-<strong>Progressive Disclosure</strong><br>
+<strong>Orthogonal Retrieval</strong><br>
-可读记忆界面保持简单，同时更深层的 facts、cases 和 skills 仍然可以被系统使用。
+按 <code>user_id</code>、<code>agent_id</code>、<code>app_id</code>、<code>project_id</code> 和 <code>session_id</code> 五维独立检索。
-</td>
-<td width="33%" valign="top">
-<strong>Modular By Design</strong><br>
-EverAlgo 负责算法；EverOS 负责运行时、持久化、在线流程和离线进化。
 </td>
 </tr>
 </table>

scripts/check_pr_title.py

@@ -0,0 +1,70 @@
+"""Validate pull request titles against the EverOS Conventional Commits policy."""
+
+from __future__ import annotations
+
+import importlib.util
+import os
+import sys
+from pathlib import Path
+from types import ModuleType
+
+_SCRIPT_DIR = Path(__file__).resolve().parent
+
+
+def _load_commit_policy() -> ModuleType:
+    policy_path = _SCRIPT_DIR / "check_commit_messages.py"
+    spec = importlib.util.spec_from_file_location("_commit_message_policy", policy_path)
+    if spec is None or spec.loader is None:
+        raise RuntimeError(f"Unable to load commit policy from {policy_path}")
+
+    module = importlib.util.module_from_spec(spec)
+    sys.modules[spec.name] = module
+    spec.loader.exec_module(module)
+    return module
+
+
+_POLICY = _load_commit_policy()
+ALLOWED_TYPES = _POLICY.ALLOWED_TYPES
+MAX_TITLE_LENGTH = _POLICY.MAX_TITLE_LENGTH
+TITLE_RE = _POLICY.TITLE_RE
+
+
+def validate_title(title: str) -> list[str]:
+    title = title.strip()
+    if not title:
+        return ["missing PR title"]
+
+    if len(title) > MAX_TITLE_LENGTH:
+        return [f"PR title is {len(title)} chars; max is {MAX_TITLE_LENGTH}: {title}"]
+
+    if not TITLE_RE.match(title):
+        allowed = ", ".join(ALLOWED_TYPES)
+        return [
+            f"invalid PR title: {title}\n"
+            "  expected: <type>[(scope)][!]: <description>\n"
+            f"  allowed types: {allowed}"
+        ]
+
+    return []
+
+
+def _title_from_args_or_env(argv: list[str]) -> str:
+    if argv:
+        return " ".join(argv)
+    return os.getenv("PR_TITLE", "")
+
+
+def main(argv: list[str] | None = None) -> int:
+    title = _title_from_args_or_env(sys.argv[1:] if argv is None else argv)
+    failures = validate_title(title)
+    if failures:
+        print("Pull request title check failed:")
+        print("\n".join(failures))
+        return 1
+
+    print("Pull request title follows Conventional Commits.")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

tests/unit/test_scripts/test_check_pr_title.py

@@ -0,0 +1,56 @@
+"""Self-tests for ``scripts/check_pr_title.py``."""
+
+from __future__ import annotations
+
+import importlib.util
+import sys
+from pathlib import Path
+
+_REPO_ROOT = Path(__file__).resolve().parents[3]
+_CHECKER_PATH = _REPO_ROOT / "scripts" / "check_pr_title.py"
+
+
+def _load_checker():
+    assert _CHECKER_PATH.exists(), "PR title checker should exist"
+    spec = importlib.util.spec_from_file_location("_pr_title_checker", _CHECKER_PATH)
+    assert spec and spec.loader
+    mod = importlib.util.module_from_spec(spec)
+    sys.modules[spec.name] = mod
+    spec.loader.exec_module(mod)
+    return mod
+
+
+def test_conventional_pr_title_is_allowed() -> None:
+    checker = _load_checker()
+
+    assert checker.validate_title("docs(readme): polish launch highlights") == []
+
+
+def test_bracketed_codex_pr_title_is_blocked() -> None:
+    checker = _load_checker()
+
+    failures = checker.validate_title("[codex] simplify README launch highlights")
+
+    assert len(failures) == 1
+    assert "invalid PR title" in failures[0]
+    assert "expected: <type>[(scope)][!]: <description>" in failures[0]
+
+
+def test_long_pr_title_is_blocked() -> None:
+    checker = _load_checker()
+    title = (
+        "docs(readme): polish launch highlights and banner with a title that is "
+        "too long"
+    )
+
+    failures = checker.validate_title(title)
+
+    assert len(failures) == 1
+    assert "max is 72" in failures[0]
+
+
+def test_main_reads_pr_title_environment(monkeypatch) -> None:
+    checker = _load_checker()
+    monkeypatch.setenv("PR_TITLE", "docs(readme): polish launch highlights")
+
+    assert checker.main([]) == 0