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

V
Visual Studio Blog
MongoDB | Blog
MongoDB | Blog
Engineering at Meta
Engineering at Meta
云风的 BLOG
云风的 BLOG
Microsoft Azure Blog
Microsoft Azure Blog
B
Blog RSS Feed
T
The Exploit Database - CXSecurity.com
P
Privacy & Cybersecurity Law Blog
Know Your Adversary
Know Your Adversary
月光博客
月光博客
I
InfoQ
阮一峰的网络日志
阮一峰的网络日志
NISL@THU
NISL@THU
爱范儿
爱范儿
S
Securelist
博客园 - 叶小钗
C
CERT Recently Published Vulnerability Notes
Recorded Future
Recorded Future
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
aimingoo的专栏
aimingoo的专栏
D
DataBreaches.Net
G
GRAHAM CLULEY
P
Proofpoint News Feed
A
About on SuperTechFans
Google DeepMind News
Google DeepMind News
C
Cyber Attacks, Cyber Crime and Cyber Security
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
T
Tor Project blog
Stack Overflow Blog
Stack Overflow Blog
T
Threat Research - Cisco Blogs
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
T
Tailwind CSS Blog
有赞技术团队
有赞技术团队
Hugging Face - Blog
Hugging Face - Blog
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
Recent Announcements
Recent Announcements
P
Proofpoint News Feed
The GitHub Blog
The GitHub Blog
The Cloudflare Blog
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
Last Week in AI
Last Week in AI
Y
Y Combinator Blog
Jina AI
Jina AI
大猫的无限游戏
大猫的无限游戏
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
罗磊的独立博客
博客园 - 【当耐特】
H
Help Net Security
F
Fortinet All Blogs
T
The Blog of Author Tim Ferriss

Hacker News: Show HN

PurrrrrFocus: Pomodoro Timer App - App Store Workflow Engine — Multi-Step Orchestration for Bun RapidPhoto: Pro Photo Editor App - App Store GitHub - DheerG/swarms: Achieve extraordinary results with claude code across a variety of tasks SPICE simulation → oscilloscope → verification with Claude Code — Lucas Gerads Show HN: VCoding – A 5 MB native Windows IDE with no dynamic dependencies Show HN: LLMs don't hallucinate because they're bad at math, it's the format GitHub - Agent-FM/agentfm-core: AgentFM is a peer-to-peer network that turns everyday computers into a decentralized AI supercomputer. AgentFM lets you run massive AI workloads directly across a global mesh of idle CPUs and GPUs. Show HN: Tracking Top US Science Olympiad Alumni over Last 25 Years GitHub - Potarix/agent-hub: One place to talk to all your agents Show HN: Runtime security for AI agents(injection,tool abuse, data exfiltration) GitHub - dubeyKartikay/lazyspotify: Terminal Spotify client for macOS and Linux GitHub - the-banana-tool/king-louie: Easy to use GUI Personal AI Assistant. Win/Linux/Mac. Show HN I made my vacation rental bookable by AI agents–no Airbnb, 0% commission GitHub - basteez/jsf-autoreload: maven plugin to enable hot reload on jsf projects uvm32/hosts/host-gdbstub at main · ringtailsoftware/uvm32 GitHub - labsai/EDDI: Config-driven engine that turns JSON into production-grade AI agents. Multi-agent orchestration, 12+ LLM providers, MCP/A2A protocols, RAG, persistent memory, and enterprise compliance (EU AI Act, GDPR, HIPAA). Built on Quarkus. GitHub - glitchnsec/fortyone-oss: AI Executive Assistant Platform Quickstart | Alien GitHub - muxshed/shed: One stream in, or many. Every destination, simultaneously. No cloud middleman, no per-channel fees, no limits. GitHub - ocrbase-hq/ocrbase: 📄 PDF/IMG ->.MD/JSON Document OCR API for PaddleOCR and GLMOCR. Self-hostable. GitHub - impactjo/home-memory: MCP server that lets your AI assistant remember everything about your home. GitHub - Sets88/dbcls: DbCls is a powerful terminal database client that supports various databases GitHub - neptun2000/heor-agent-mcp GitHub - SeanFDZ/macmind: Single-layer transformer in HyperTalk for the classic Macintosh RollQuation: Math Puzzles - Apps on Google Play GitHub - dropbox/witchcraft Show HN: Agent-cache – Multi-tier LLM/tool/session caching for Valkey and Redis GitHub - opentalon/opentalon: OpenTalon is an open-source platform built from the ground up in Go as a robust alternative to OpenClaw LinkedIn™ 职位抓取工具 - Chrome 应用商店 GitHub - EdoardoBambini/Agent-Armor-Iaga: AI agents are getting tool access — shell, file system, databases, APIs, secrets. But **nobody is governing what they actually do with it**. Frameworks like LangChain, CrewAI, AutoGen, and Claude Code give agents the power to execute. Agent Armor gives you the power to control, audit, and approve every single action before it happens. HN Vibes — Week 15, Apr 7–13 2026 GitHub - chojs23/ec: Easy terminal-native 3-way git mergetool vim-like workflow GitHub - SethPyle376/hiraeth: Local AWS emulator focused on fast integration testing, with SQS support, SQLite-backed state, and a debug-friendly web UI. GitHub - JakOb-dotcom/cloud-sandbox-security-analysis: Technical analysis and Proof of Concept (PoC) regarding environment variable exfiltration in containerized cloud sandboxes via side-channel data leaks. Springboards - Flint Alpha Show HN: A simpler coding agent harness GitHub - audiodude/sudomake-friends GitHub - 256thFission/mini-mythos: OSS clone of Anthropic’s Mythos harness to locate C/C++ memory vulnerabilities Show HN: OpenParallax: OS-level privilege separation for AI agent execution Hacker News Sorted - Chrome 应用商店 Show HN: How to Install Docker on Ubuntu 24.04 LTS: Complete 2026 Guide GitHub - himanshudongre/smriti GitHub - sverrirsig/claude-control: macOS desktop dashboard for monitoring and managing multiple Claude Code sessions GitHub - ory/dockertest: Write better integration tests! Dockertest helps you boot up ephermal docker images for your Go tests with minimal work. Chiral - Chrome 应用商店 Show HN: Two Claudes collaborating through shared memory on a $100 mini-PC GitHub - pmichaillat/latex-cv: Minimalist LaTeX template for academic CVs GitHub - oguzbilgic/posse: A web UI for Anthropic Managed Agents. GitHub - sshiraz/depsly: Dependency risk analysis tool for npm packages ABI Add safari/agent-harness — Safari browser automation via safari-mcp by achiya-automation · Pull Request #212 · HKUDS/CLI-Anything GitHub - Halfblood-Prince/trustcheck: Verify PyPI package attestations and improve Python supply-chain security GitHub - oguzbilgic/kern-ai: Agents that do the work and show it. GitHub - bruits/satteri: High-performance Markdown and MDX processing for the JavaScript ecosystem GitHub - tylergibbs1/feedstock: High-performance web crawler and scraper for TypeScript, powered by Bun and Playwright GitHub - Grimm67123/grimmbot: The self-improving sandboxed and open-source AI agent. With persistent memory and scheduling. GitHub - whitevanillaskies/whitebloom: Local whiteboard that blooms. GitHub - hwdsl2/docker-whisper: Docker image for a self-hosted Whisper speech-to-text server with speaker diarization and OpenAI-compatible transcription and translation APIs. Powered by faster-whisper. Supports all Whisper models, NVIDIA GPU (CUDA) acceleration, JSON/SRT/VTT output, SSE streaming, offline mode, and multi-arch (amd64, arm64). GitHub - yisding/reviewwiggum GitHub - MarwanAlsoltany/serrors: Structured errors for Go: sentinel hierarchies, typed data, custom formatting, and slog integration. GitHub - soatok/age-php GitHub - Luthiraa/markitme GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits GitHub - tombedor/excalicharts GitHub - wh1le/excalidraw-edit: Open and edit .excalidraw files from the terminal. Offline, auto-saves to disk. MalExt Sentry - Malicious Extension Scanner - Chrome 应用商店 GitHub - syi0808/asciianimesvg: Generate animated ASCII art SVGs from text. CLI, Rust library, WASM, and web editor. GitHub - zaina-ml/ml_forge: A visual-based graph node editor for training computer vision models. GitHub - anakin87/llm-rl-environments-lil-course: 🌱 A little course on Reinforcement Learning Environments for evaluating and training Language Models GitHub - takaakit/superpowers-uml: Superpowers-UML modifies Superpowers to ensure a software development workflow in which AI agents design through UML modeling. AdriByte Studio - Sviluppo Web e Soluzioni Digitali GitHub - chouligi/angel-copilot: Your personalized Angel Investment Advisor Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 GitHub - agenteractai/lodmem: Level Of Detail Context Management for Agents GitHub - ostefani/subnetlens: A fast, concurrent network scanner with a TUI and plain-text CLI, built in Go. It discovers live hosts on your network, scans their open ports, resolves hostnames, and fingerprints operating systems—delivered. Cyber Pulse: Agentic Intel - Apps on Google Play Whisper API: Self-Hostable Speech to Text Transcription The Agent-Web Protocol Stack: A Research Thesis GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Show HN: Provepy – A Python decorator that proves your code using Lean and LLMs Show HN: Pardonned.com – A searchable database of US Pardons GitHub - patrickdappollonio/dux: Dux is a terminal UI that lets you run multiple AI coding agents side by side, each in its own git worktree, with full companion terminals, macros, commit generation, and a command palette that knows more tricks than you do. kMC Crystal Simulator Show HN: HyperFlow – A self-improving agent framework built on LangGraph GitHub - stef41/vibescore: 🎵 Grade your vibe-coded project. One command, instant letter grade across security, quality, dependencies, and testing. GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. imgur.com GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. GitHub - nowork-studio/toprank: Open-source Claude Code skills for SEO, SEM, Google Ads GitHub - tacomanator/sash: Lightweight macOS menu bar app for reliably cycling through windows of the current application. Appents | Social Media Management for Product-First Teams GitHub - pnhoang/youtube-spam-blocker: Automatically detects and hides spam messages in YouTube Live chat. Set rate limits, keyword filters, and block repeat offenders. GitHub - decisionnode/DecisionNode: CLI + Local MCP - A shared structured memory store across Claude Code, Cursor, Windsurf, Antigravity, and every MCP client. Semantically queryable. GitHub - AvaCodeSolutions/django-email-learning: An open source Django app for creating email-based learning platforms with IMAP integration and React frontend components. The $100K Gap in Kubernetes Security Tooling Function Calling Harness: From 6.75% to 100%
GitHub - kernalix7/AIPS: AI Project Setup
kernalix7 · 2026-05-19 · via Hacker News: Show HN

Status: v7.0.0 stable — shipped 2026-05-19

AIPS v7.0 is the current, supported release. The 11 /aips:* slash commands, hybrid global-first layout, 4 bundled plugins, and 3-line statusline are all in place and tested on Linux, macOS, and Git Bash / WSL. If you're already running v6.0, migration is one command (/aips:upgrade --to v7.0) and the strict mode result is byte-identical to a fresh v7.0 install.


Table of contents


Why this exists

You've been here before:

  • Cloned a fresh repo, then spent an hour re-wiring CLAUDE.md, .cursorrules, AGENTS.md, the same 4 hooks, the same 9 backup scripts.
  • Lost a 2-hour session to a /clear because nobody wrote a handoff.
  • Re-installed RTK and caveman on a new laptop and forgot which order matters.
  • Watched the AI burn 25k tokens reading a setup doc before doing actual work.
  • Renamed a project directory and watched session history evaporate.

AIPS v7.0 fixes all of that with a single principle: state that should be global, lives global. State that should be local, lives local. Nothing duplicated.

Without AIPS With AIPS v7
Hand-write CLAUDE.md from scratch — hours /aips:init — ~30 seconds
Hooks, agents, statusline copied per project Once per machine in ~/.claude/, every project inherits
Session lost on crash / compact / /clear 3-tier autosave (current / handoff / recovery) reloaded on next start
Memory tied to one repo path Path-hash-keyed in ~/.claude/projects/, survives renames via /aips:rebind
22-line .gitignore block in every repo One block in ~/.config/git/ignore, every repo gets it free
AI reads a 7,600-line setup doc each session AI reads a ~150-line CLAUDE.md; the rest is shell-level deterministic

Quick start

Once per machine — runs for every project you'll ever open:

curl -fsSL https://raw.githubusercontent.com/kernalix7/AIPS/main/install.sh | bash

What install.sh does:

  1. Registers the AIPS marketplace in ~/.claude/
  2. Installs / updates 4 bundled plugins:
    • codex-plugin-cc/codex:* commands (Claude ↔ Codex relay)
    • caveman — ultra-terse output style + /caveman* commands
    • agentmemory + systemd unit — file-backed memory + auto-backup
    • RTK (Rust Token Killer) — 60–90% token savings CLI proxy
  3. Drops hooks / agents / commands / skills / output-styles / statusline into ~/.claude/
  4. Globalizes toolkit scripts (~/.local/bin/aips-*) and the AIPS gitignore block (~/.config/git/ignore)

Per project:

cd my-project
claude
> /aips:init

/aips:init auto-detects 4 cases:

Case Trigger Action
A. Fresh No .priv-storage/, no v5.x .md Fresh init (~30 sec)
B. v5.x migrate Root AI_PROJECT_SETUP.md detected Confirm → backup → cleanup → v7.0 init
C. Re-init (idempotent) .priv-storage/ v7.0 marker present Drift repair, no destructive ops
D. Repair Broken / partial state detected Repair mode with selective rebuild

Done. Every artifact is gitignored — your git history stays clean. Verify any time with /aips:health.


What lands where

v7.0 keeps a clean split: smart globals once per machine, minimum-viable per-project state per repo. Nothing is duplicated.

Global — ~/.claude/ + ~/.local/bin/ + ~/.config/git/ignore

Installed once by install.sh. Every project inherits.

Category Contents
Plugins 4 bundled: codex-plugin-cc, caveman, agentmemory (+ systemd unit), RTK
Hooks 5: PreToolUse, PostToolUse, SessionStart, PreCompact, Stop
Agents 3 templates: tech-lead, explorer, code-reviewer
Commands 15: 11 /aips:* + bundled plugin commands
Skills On-demand knowledge modules (caveman, codex, etc.)
Output styles terse (default), caveman/full, caveman/ultra, etc.
Statusline 3-line multi-line (preview below)
Binaries RTK Rust binary (~/.local/bin/rtk), aips-* toolkit symlinks
Daemons agentmemory systemd unit (user-level)
Sessions mirror ~/.claude/sessions/{path-hash}/ — survives project renames
Memory store ~/.claude/projects/{path-encoded}/memory/ — single source of truth
Gitignore block ~/.config/git/ignore — 22 entries, zero per-repo noise

Per-project — .priv-storage/ (/aips:init)

your-project/
|-- CLAUDE.md              -> .priv-storage/CLAUDE.md  (symlink)
|-- AGENTS.md              -> .priv-storage/CLAUDE.md  (symlink, Codex / Copilot)
|-- .cursorrules           -> .priv-storage/CLAUDE.md  (symlink, Cursor)
|-- WORK_STATUS.md         -> .priv-storage/WORK_STATUS.md
|
|-- .priv-storage/         [gitignored] per-project AI state
|   |-- CLAUDE.md                   # Sections 1-7 + 11 only, ~150 lines
|   |-- CLAUDE.local.md             # per-developer overrides
|   |-- WORK_STATUS.md              # team-shared task state
|   |-- sessions/                   # fast-write buffer; mirrored globally
|   |-- agents/                     # tech-lead.md + per-team agents
|   `-- .mcp.json                   # project-specific MCP server registry
|
`-- tmp-igbkp/             [gitignored] encrypted backup outputs only
    `-- (archives produced by aips-archive, aips-restore, aips-purge-history)

Only WORK_STATUS.md, GitHub standard files, the docs/ Korean mirrors, and .github/ get committed. Everything in .priv-storage/ and tmp-igbkp/ is gitignored on purpose. CLAUDE.md keeps Sections 1–7 + 11 (~150 lines) — Sections 8/9/10/12/13 moved into the global plugin / skill / hook layer.

Why this split

Item Lives Why
Toolkit scripts (aips-archive, aips-restore, etc.) global One canonical copy, no per-project drift, single PATH lookup
Sessions mirror global (project keeps fast-write buffer) Survives project-dir renames, cross-machine via single dir
Memory store global only Single source of truth, no dual-write drift
.gitignore AIPS block global Zero-noise project gitignore; works across all repos automatically
CLAUDE.md Sections 1–7 + 11 per-project Team-shared project rules; multi-tool guarantee
WORK_STATUS.md per-project Team-shared task state — must live in the repo
.mcp.json per-project Project-specific MCP server registry
tech-lead.md + team agents per-project Per-project team composition
tmp-igbkp/ backup outputs per-project Encrypted archives belong with the project they back up

Lifecycle

[1. Global install] (once per machine)
  curl ... | bash
  - Register AIPS marketplace → ~/.claude/
  - Install/update 4 bundled plugins
  - Drop hooks/agents/commands/skills/output-styles/statusline
  - Globalize toolkit (~/.local/bin/aips-*) + gitignore block
  - Install RTK binary, agentmemory systemd unit
        |
        v
[2. Project init] (once per project, ~30 sec)
  cd project && claude → /aips:init
  - Auto-branch A/B/C/D
  - Create .priv-storage/, drop 3 symlinks
  - CLAUDE.md Sections 1-7 + 11 (~150 lines)
  - Register project with global sessions mirror + memory store
        |
        v
[3. Normal session]
  - SessionStart hook auto-injects prior handoff + current.md tail
  - AI reads only CLAUDE.md (~150 lines) — no 7,600-line .md
  - PreToolUse blocks dangerous commands + oversized reads
  - PostToolUse appends current.md + dual-writes to agentmemory
  - Stop hook writes handoff-{date}.md
  - PreCompact hook writes recovery.md (best-effort)
  - 3-line statusline updates live: project / windows / savings
        |
        v
[4. Crash / rate-limit / /clear / rename]
  - Next session, SessionStart auto-loads handoff + recovery
  - AI resumes prior state without asking "where were we?"
  - Renamed project? `/aips:rebind <old-path>` rewires globals
        |
        v
[5. Updates] (trigger: /aips:update)
  - marketplace pull → update 4 bundled plugins
  - Global hooks/commands auto-sync
  - Projects untouched (re-init via /aips:init if needed)

Statusline

3-line multi-line. Three glances tell you everything: what you're working on, how much budget you have left, and how much you've saved.

project [main*3] wip:2 | opus-4.7 | ctx:8%(15.5k/200k) | cache:71%
5h:8% ↻2h11m ∅1h23m | wk:12% ↻4d18h ∅2d4h
🦴cv:75%/full | 🧠am:40%/127 | 💰rtk:34% | 🤖cdx:55%/3runs | 💯Σ:95%
Line Shows
1 Project name, git branch [commits since base], wip count, model, context usage, prompt cache hit rate
2 5-hour window usage + time until reset () + projected empty ETA at current burn rate (); weekly window same
3 caveman savings/intensity, agentmemory hit rate / observation count, RTK savings, codex delegation rate / runs, cumulative savings (Σ)

= time until reset · = projected empty ETA at current burn rate · Σ = cumulative savings (Σ = 1 − Π(1 − SAVED_i / 100)).


Slash commands

AIPS native — 11 commands

Command Action
⚙️ /aips:init Auto-branched init (fresh / migrate / re-init / repair)
🔄 /aips:update marketplace pull + bundled plugin update
🩺 /aips:health verify-setup.sh + smoke-test-hooks.sh
🗑️ /aips:uninstall Safe removal, global + project
📊 /aips:status Current task + recent activity summary
📜 /aips:migrate-from-md Explicit v5.x .md migration (manual case-B trigger)
⬆️ /aips:upgrade Version upgrade — supports --to v7.0 for v6.0 → v7.0 hybrid migration
🔧 /aips:repair Repair broken state (manual case-D trigger)
♻️ /aips:reset Reset project init (with backup)
🔗 /aips:rebind <old-path> Rebind globalized state when a project moves or is renamed
🔍 /aips:scope Diagnose what is globalized vs per-project; flag drift or orphans

Bundled plugin commands

Plugin Commands
codex-plugin-cc /codex:brief, /codex:review, /codex:fix, /codex:relay-status
caveman /caveman, /caveman:lite, /caveman:ultra, /caveman:wenyan-*
agentmemory /am:save, /am:recall, /am:reflect, /am:consolidate, /am:sessions
RTK Hook-based auto — no explicit command (rtk gain from shell)

Legacy custom /codex-brief, /codex-review, /codex-fix, /codex-relay-status from v5.x were retired in v6.0. Use /codex:* from the official codex-plugin-cc instead.


Migration

From v5.x — /aips:init handles it

cd existing-v5-project
claude
> /aips:init
# → Detects root AI_PROJECT_SETUP.md (v5.x)
# → "v5.x install detected. Migrate to v7.0? [y/N]"
# → On y:
#   1. Full backup to .priv-storage/v5-backup/
#   2. Reduce 7,600-line AI_PROJECT_SETUP.md → 30-line DEPRECATED redirect
#   3. Remove custom /codex-* (4 commands) — replaced by codex-plugin-cc
#   4. Slim CLAUDE.md to Sections 1-7 + 11 only (~150 lines)
#   5. Register with global sessions mirror + memory store
#   6. Write v7.0 marker

Or trigger it explicitly: /aips:migrate-from-md.

From v6.0 — /aips:upgrade --to v7.0

cd existing-v6-project
claude
> /aips:upgrade --to v7.0
# → Strict mode (default): result is identical to a fresh v7.0 install.
#   per-project tmp-igbkp/*.sh and sessions/*.md are purged after their
#   global counterparts are verified. Full backup at
#   tmp-igbkp/upgrade-v7-backup-{ts}/ is always taken first.
# → Pass --keep-local-fallback to retain both as fallback (lenient).

What v7.0 changed vs v6.0

Item v6.0 v7.0
Toolkit scripts tmp-igbkp/*.sh copied per project ~/.local/bin/aips-* symlinks (one canonical copy)
Sessions .priv-storage/sessions/ only per-project buffer + ~/.claude/sessions/{path-hash}/ mirror
Memory per-project + global (dual-write) global only — ~/.claude/projects/{path-encoded}/memory/
Gitignore 22 entries in every .gitignore one block in ~/.config/git/ignore
New commands /aips:rebind, /aips:scope, /aips:upgrade --to v7.0

Supported AI tools

AIPS is built for Claude Code first. Other tools get policy-only support via CLAUDE.md / AGENTS.md / .cursorrules. Full plugin parity for them is on the roadmap.

Tier 1 — Primary / Full

Tool Min version Reads Features
Claude Code (CLI) 2.0+ CLAUDE.md Full plugin install, 11 /aips:* commands, 5 hooks, 3-line statusline, output styles, 4 bundled plugins

Tier 2 — Partial (policy-only)

Tool Min version Reads Features
ChatGPT Codex CLI 0.10+ AGENTS.mdCLAUDE.md Rules only (no hooks / slash / statusline)
Cursor 0.40+ .cursorrulesCLAUDE.md Rules only
GitHub Copilot 1.150+ AGENTS.md Rules only
claude.ai (web) current CLAUDE.md manual upload Rules only
Any MCP-aware tool depends .mcp.json only

Policy-only = rules enforced through prompt content. No kernel-level blocking, no hooks, no slash commands, no statusline — but the rule file is read by the AI and followed.

Tier 3 — Full support TBD

Full plugin-like parity (hooks, slash commands, statusline, bundled-plugin stack) for Codex / Cursor / Copilot is roadmap. Track progress in Roadmap or open an issue to upvote.


Comparison

AIPS v7.0 .cursorrules only Hand-written CLAUDE.md Nothing
Per-project setup time ~30 sec instant hours 0
Deterministic (no AI interpretation) yes yes yes n/a
Single source of truth across tools yes Cursor only Claude only no
One-line global install yes no no n/a
Safety hooks (kernel-level) yes no manual no
Session resume on crash / /clear yes (3-tier) no manual no
3-line statusline (incl. savings) yes no no no
Cross-AI relay (Claude ↔ Codex) yes (plugin) no no no
Token savings (RTK + caveman + agentmemory) yes no no no
Upstream self-update yes (/aips:update) no no n/a
Survives project rename yes (/aips:rebind) no manual no
Zero per-repo gitignore noise yes (global ignore) yes no yes
Linux / macOS / Windows yes yes yes n/a

Documentation

Document What's inside
install.sh Global install script
docs/README.ko.md This README in Korean
AI_PROJECT_SETUP.md v5.2 archive (30-line DEPRECATED stub for downstream raw-URL compatibility)
CONTRIBUTING.md · 한국어 Dev setup, version-bump checklist, PR conventions
SECURITY.md · 한국어 Disclosure process, secret-guard patterns
CODE_OF_CONDUCT.md · 한국어 Contributor Covenant v2.1
CHANGELOG.md · 한국어 Full version history

FAQ

Why a 150-line CLAUDE.md instead of dumping everything in?

Because your AI loads it every session. Token discipline = your wallet. The other 12 sections of project policy live in global hooks, skills, and output-styles — they execute deterministically without the AI re-reading them.

Do I still have to upgrade from v6.0?

No, v6.0 keeps working. v7.0 is opt-in and non-breaking. Run /aips:upgrade --to v7.0 when you want a single canonical toolkit, sessions mirror, global gitignore, and single-source memory.

What if I rename or move my project after v7.0 init?

/aips:rebind <old-path> from inside the project's new location. It rewrites the path-hash-keyed globals (~/.claude/sessions/{path-hash}/, memory mappings) so sessions and memory continue resolving to the same project.

How do I see what's globalized vs per-project?

/aips:scope. It prints which artifacts live where for the current project, and flags drift (per-project sessions buffer ahead of global mirror) or orphans (global state for a project dir that no longer exists).

What does install.sh touch on my system?

~/.claude/, ~/.local/bin/, ~/.config/git/ignore, and a user-level systemd unit (agentmemory). It does not touch system-wide directories (/usr/local/, /etc/). Uninstall via /aips:uninstall.

Why bundle 4 plugins instead of letting me pick?

Each one is orthogonal — codex-plugin-cc (relay), caveman (output compression), agentmemory (persistent memory), RTK (CLI token savings). Installed separately, hook order and statusline fallbacks become painful. Bundling guarantees cross-plugin sync; you can still disable any of them via Claude Code's plugin settings.

Does it work offline?

Install / update need the network (marketplace pull, RTK binary fetch). Normal sessions and /aips:init work offline — every artifact is read from the local plugin store.

How do I sync across multiple machines?

agentmemory writes project memory to ~/.claude/projects/{path-encoded}/memory/. On a new machine: install.sh/aips:init in the project → rsync the memory dir over. Sessions mirror works the same way under ~/.claude/sessions/.

Windows?

Git Bash, WSL, and MSYS2 work. install.sh is bash, hooks are bash, so native PowerShell is not supported. WSL is recommended.

Multiple AI tools in the same project?

Yes. Claude Code reads CLAUDE.md, Codex / Copilot read AGENTS.md, Cursor reads .cursorrules — all three are symlinks to the same .priv-storage/CLAUDE.md, so updates are atomic. Hooks, statusline, and slash commands are Claude Code-only.

Can I skip the marketplace and just clone the repo?

Yes. Clone the repo and run install.sh from the local path — the marketplace step falls back to pointing at the local path. Useful for forks and air-gapped environments.

Found a bug or want a feature.

Open an issue or PR at https://github.com/kernalix7/AIPS. See CONTRIBUTING.md.


Roadmap

  • v7.0.0 (stable, 2026-05-19) — Hybrid global-first: globalized toolkit / sessions / memory / gitignore, 3 new /aips:* commands (upgrade --to v7.0, rebind, scope), strict-by-default non-breaking migration from v6.0
  • v7.1 — agentmemory deeper integration: cross-project workflow recommendations, shared lesson surfaces visible in /am:recall
  • v7.2/aips:rebind UX: auto-detect moved projects via path-hash heuristics; one-command recovery without specifying <old-path>
  • v7.3 — Tier 2 → Tier 1 promotion experiments: prototype hook / statusline support for Codex CLI via wrapper shim
  • v8.0 (candidate) — Team-shared globals via opt-in cloud sync, or full plugin-marketplace publishing for third-party AIPS extensions

See CHANGELOG.md for shipped versions.


Star history

Star History Chart

Support

If AIPS saved you setup time:

Ko-fi Fairy

Ko-fi handles international cards and PayPal; fairy.hada.io is a Korean tipping platform. Bug reports, PRs, and stars on the repo are equally appreciated and free.


License

MIT — Kim DaeHyun (kernalix7@kodenet.io)