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
MyScale Blog
MyScale Blog
博客园 - Franky
The Cloudflare Blog
IT之家
IT之家
Blog — PlanetScale
Blog — PlanetScale
博客园 - 聂微东
WordPress大学
WordPress大学
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
T
The Blog of Author Tim Ferriss
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
罗磊的独立博客
Google DeepMind News
Google DeepMind News
P
Proofpoint News Feed
Martin Fowler
Martin Fowler
aimingoo的专栏
aimingoo的专栏
J
Java Code Geeks
腾讯CDC
雷峰网
雷峰网
Microsoft Azure Blog
Microsoft Azure Blog
G
Google Developers Blog
博客园 - 【当耐特】
美团技术团队
云风的 BLOG
云风的 BLOG

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders
GitHub - Vortrix5/sifty: Windows maintenance CLI + TUI th...
vortrix5 · 2026-06-13 · via Hacker News - Newest: "AI"

Sifty

Sift the junk from the keep.

Python 3.11+ Windows License: MIT CI

Clean junk, analyze disks, find duplicates, manage apps and startup programs, apply updates, prune dev artifacts and git worktrees, and organize files, from a scriptable CLI or a full-screen terminal UI. The optional AI assistant runs locally via Ollama: nothing leaves your machine, and it only ever sees file metadata (names, sizes, paths), never file contents.

Sifty demo

Safety first

Sifty deletes files and changes system state, so it is built to be hard to misuse:

  • Dry-run by default. Every destructive command previews what it would do; real changes need an explicit --apply.
  • Recycle Bin, never permanent delete. All removals go through one trash() function backed by Send2Trash, and sifty undo restores the last clean.
  • Protected paths. C:\Windows, Program Files, ProgramData, the drive root and your profile root are refused even with --apply --yes.
  • Audit log. Every applied deletion is recorded in %APPDATA%\sifty\audit.log.
  • The AI never deletes anything. It is advisory; high-risk tool calls always require your approval.

How it compares

Feature Sifty CCleaner Revo Uninstaller WinDirStat
Junk / cache cleaning ✅ 11+ categories
Disk usage analysis ✅ top-N + volumes ✅ treemap
Duplicate finder ✅ SHA-256, NTFS-aware ✅ (paid)
App uninstall + leftover scan ✅ winget + leftovers ✅ + leftovers
App updates ✅ via winget ✅ (paid)
Startup manager ✅ reversible
Dev artifact purge (node_modules, …)
Git worktree / WSL2 VHD cleanup
Local AI assistant ✅ Ollama
Scriptable (JSON output)
Recycle Bin + undo for everything n/a
Price Free, MIT Freemium Freemium Free

Sifty is built developer-first: everything is scriptable, the engine is a reusable Python library, and it cleans the things developer machines actually accumulate (build artifacts, orphaned worktrees, bloated WSL2 disks).

Install

pipx install sifty       # recommended (isolated); or: pip install sifty
scoop bucket add sifty https://github.com/Vortrix5/scoop-bucket; scoop install sifty
winget install Vortrix5.Sifty            # once the winget-pkgs PR is merged

sifty doctor             # check admin rights, winget, Ollama

Prefer no Python install? Download the standalone sifty.exe from the latest release.

For development:

python -m venv .venv
.\.venv\Scripts\python.exe -m pip install -e ".[dev]"
.\.venv\Scripts\python.exe -m pytest -q

Usage

sifty checkup                # one read-only scan of everything: junk, updates,
                             # orphans, stale files, disk space, startup
sifty tui                    # the full-screen interactive app

# Junk
sifty junk scan              # show reclaimable space per category
sifty junk clean             # preview removal (dry-run)
sifty junk clean --apply     # send junk to the Recycle Bin (asks first)

# Disk
sifty disk volumes           # used/free/total per volume
sifty disk analyze C:\Users  # biggest folders/files under a path
sifty disk duplicates D:\    # find duplicate files and wasted space

# Apps & updates
sifty apps list --by-size    # installed apps, largest first
sifty apps orphans           # broken uninstall entries in the registry
sifty apps uninstall "App"   # uninstall via winget (preview, then --apply)
sifty apps leftovers "App"   # what the uninstaller left behind (then --apply)
sifty update check           # available updates (winget)
sifty update apply           # upgrade everything (asks first)

# Developer cleanup
sifty purge clean            # node_modules, dist, __pycache__, target, …
sifty cleanup duplicates D:\Photos   # de-duplicate (keeps one copy each)
sifty cleanup large C:\Users\you     # biggest files under a path
sifty cleanup stale --days 180       # old items in Downloads

# Startup & services
sifty startup list                   # startup programs (enabled/disabled)
sifty startup disable "Spotify"      # reversible (sifty startup enable …)
sifty services list                  # curated optional services + state
sifty --admin services disable DiagTrack   # toggle one (needs admin)

# History & undo
sifty history                # what was cleaned + total space reclaimed
sifty undo                   # restore the most recent clean from the Recycle Bin

# Organize files
sifty organize preview C:\Users\you\Downloads --by type
sifty organize apply   C:\Users\you\Downloads --by date
sifty organize undo          # put the last organize's files back

# Configuration
sifty config                 # all settings + which ones you've overridden
sifty config set ai.model "llama3.2:3b"
sifty config edit            # open config.toml in your editor

# AI (requires Ollama running)
sifty ai status
sifty ai ask "what can I safely delete on my C drive?" --path C:\

# Scripting: JSON output on read-only commands (auto-enabled when piped)
sifty --json checkup
sifty --json disk volumes
sifty --json apps list --by-size

Some operations (Windows temp, update cache, certain uninstalls) need an Administrator terminal. sifty doctor tells you if you're elevated, and sifty --admin <cmd> relaunches elevated via UAC.

The TUI

sifty tui opens a full-screen app with a seven-section sidebar (Home, Clean, Disk, Apps, Monitor, Reports, AI):

  • Home: volume gauges and a Run checkup button that scans everything at once; findings come with buttons that fix them right there (clean junk, clean stale downloads, apply updates, each behind a confirm).
  • Clean: Junk / Purge / Optimize / Smart cleanup under one roof (tabs).
  • Apps: Installed / Updates / Startup / Services, with fuzzy filter, sorting, bulk uninstall, and an automatic leftover scan after uninstalling.
  • AI: an agentic chat where proposed tool runs show Run/Skip buttons inline in the conversation, and scan results carry follow-up action buttons.

Press Ctrl+P for the command palette (jump to any screen), F2 to elevate, Space to mark rows for bulk actions. The Reports screen shows space reclaimed over time with an Undo last clean button.

AI setup (optional)

  1. Install Ollama and start it.
  2. Pull the configured model: ollama pull qwen2.5:3b.
  3. sifty ai status should report "running".

Configure everything with the sifty config command, no need to hand-edit files:

sifty config                                  # show all settings + your overrides
sifty config set ai.model "llama3.2:3b"       # use a different local model
sifty config set ai.host "http://localhost:11434"
sifty config set safety.extra_protected_paths '["D:\\Important"]'
sifty config set junk.include_downloads_installers true
sifty config edit                             # or open config.toml in your editor

Settings live in %APPDATA%\sifty\config.toml; sifty config set only writes the keys you change, so defaults keep flowing through on upgrades.

Architecture

Layered: thin frontends over a reusable engine, OS specifics quarantined:

src/sifty/
├── cli/               # Typer entry point + one thin command module per group
├── tui/               # Textual full-screen app (views call core, not cli)
├── core/              # the engine: junk, disk, apps, updates, cleanup,
│   │                  # startup, services, organize, checkup, history, …
│   └── safety.py      # ★ protected paths, dry-run guard, trash(), audit log
├── windows/           # OS primitives: winget, registry, UAC, Recycle Bin, DISM
├── ai/                # Ollama client, advisor prompts, agentic tool loop
└── infra/             # TOML config + rotating diagnostics log

Frontends depend on core; core depends on windows/infra; nothing imports upward. A GUI could call the same engine functions (junk.scan, disk.find_duplicates, checkup.run_checkup) without a rewrite. See docs/ARCHITECTURE.md for the design rationale.

Tests

.\.venv\Scripts\python.exe -m pytest -q    # 160+ tests, ~20 s

The safety guardrails are the most heavily tested code in the repo; the Windows environment is mocked so the suite also runs on CI. See CONTRIBUTING.md to get involved.

License

MIT © Amine Zouaoui