惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

月光博客
月光博客
罗磊的独立博客
The GitHub Blog
The GitHub Blog
V
V2EX
Last Week in AI
Last Week in AI
博客园 - 聂微东
MyScale Blog
MyScale Blog
美团技术团队
L
LangChain Blog
博客园 - Franky
腾讯CDC
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
博客园_首页
S
SegmentFault 最新的问题
爱范儿
爱范儿
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
Stack Overflow Blog
Stack Overflow Blog
量子位
小众软件
小众软件
宝玉的分享
宝玉的分享
J
Java Code Geeks
Google DeepMind News
Google DeepMind News
D
Docker
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work Top 15 Reinforcement Learning Questions That Will Appear in Exams The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026
clerk: Auto-Summarize Your Claude Code Sessions
Vulcan Shen · 2026-04-17 · via DEV Community
Cover image for clerk: Auto-Summarize Your Claude Code Sessions

Vulcan Shen

What is clerk

clerk is a CLI tool that hooks into Claude Code. Every time a session ends, it generates an incremental summary and saves it as a plain markdown file. Over time, your sessions become a searchable, organized knowledge base.

brew install vulcanshen/tap/clerk
clerk install

After that, it runs silently. No commands to remember, no habits to build.

What it solves

Claude Code has no memory across sessions. When you need to look back — recover context, search past work, or generate a report — there's nothing to work with.

You could ask Claude to re-read old transcripts, but each time it re-processes the entire raw conversation, burning tokens. Across multiple sessions and projects, that's expensive and slow.

clerk does one API call per session at the moment it ends. By the time you need a report, everything is already summarized.

What you can do with it

Weekly reports

clerk report --days 7

Reads all summaries from the past 7 days, sends them to Claude, outputs a structured report with summary, by-date, and by-project views. One command, one API call.

Context recovery

Type /clerk-resume in Claude Code. clerk returns your past summaries and transcript paths so Claude can rebuild context without you hunting for session IDs.

Search

Type /clerk-search in Claude Code. Search past work by keyword across all projects using AI semantic matching.

Daily summaries

Automatic. Every session end produces an incremental summary, organized by date and project:

~/.clerk/summary/
├── 20260414/
│   ├── my-api-server.md
│   └── frontend-app.md
└── 20260418/
    └── my-api-server.md

How it works

Session ends → clerk feed (background) → read transcript → call claude -p → save summary + index

  • Hooks into Claude Code via SessionStart/SessionEnd hooks
  • Cursor tracking — only processes new messages since last run
  • MCP server for /clerk-resume and /clerk-search integration
  • Single Go binary, no dependencies beyond Claude Code

Your data, your tools

All output is plain markdown with YAML frontmatter. No proprietary format, no lock-in.

Your files work with any text editor, Obsidian, Notion, grep, or your own scripts. If you uninstall clerk and Claude Code, your summaries remain — organized, searchable, and linked.

Everything runs locally. No remote services, no accounts, no data leaving your machine.

Install

# Homebrew
brew install vulcanshen/tap/clerk

# Or install script
curl -fsSL https://raw.githubusercontent.com/vulcanshen/clerk/main/install.sh | sh

# Or from source
go install github.com/vulcanshen/clerk@latest

Then run clerk install to set up hooks, MCP server, and skills.

Cross-platform: macOS, Linux, Windows.


GitHub: github.com/vulcanshen/clerk