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

腾讯CDC
N
Netflix TechBlog - Medium
Google DeepMind News
Google DeepMind News
Scott Helme
Scott Helme
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
小众软件
小众软件
月光博客
月光博客
有赞技术团队
有赞技术团队
Microsoft Security Blog
Microsoft Security Blog
爱范儿
爱范儿
WordPress大学
WordPress大学
Jina AI
Jina AI
M
MIT News - Artificial intelligence
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
阮一峰的网络日志
阮一峰的网络日志
B
Blog RSS Feed
P
Proofpoint News Feed
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
F
Fortinet All Blogs
Y
Y Combinator Blog
Microsoft Azure Blog
Microsoft Azure Blog
云风的 BLOG
云风的 BLOG
Hugging Face - Blog
Hugging Face - Blog
MongoDB | Blog
MongoDB | Blog
I
InfoQ
Vercel News
Vercel News
C
Check Point Blog
美团技术团队
V
V2EX
量子位
博客园 - 三生石上(FineUI控件)
D
DataBreaches.Net
G
Google Developers Blog
博客园_首页
J
Java Code Geeks
Recent Announcements
Recent Announcements
人人都是产品经理
人人都是产品经理
H
Help Net Security
博客园 - Franky
The GitHub Blog
The GitHub Blog
V
Visual Studio Blog
T
Tailwind CSS Blog
IT之家
IT之家
S
SegmentFault 最新的问题
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
雷峰网
雷峰网
L
LangChain Blog
博客园 - 司徒正美
T
The Blog of Author Tim Ferriss
H
Hackread – Cybersecurity News, Data Breaches, AI and More

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 - Giammarco-Ferranti/deja: Predictive inline shell autosuggestions for zsh - Go daemon, no TUI, no sync
giammiferr · 2026-06-07 · via Hacker News: Show HN

Deja mascot

Predictive ghost-text autosuggestions for zsh — smarter than history, lighter than a plugin.

Latest release License Go Report Card


Deja is a smarter replacement for zsh-autosuggestions. Instead of only surfacing commands that start with what you've typed, Deja uses fuzzy matching, directory awareness, and command sequence prediction to suggest what you actually want to run — as inline ghost text, after every keystroke, with zero latency.

No account. No sync server. No TUI. Just ghost text that knows where you are.

Deja in action


Features

  • Fuzzy matching — suggests commands even when you skip letters or mix up order
  • Directory awareness — commands you run in ~/projects/foo rank higher when you're in ~/projects/foo
  • Sequence prediction — knows that you usually run make test after make build
  • Frecency scoring — blends frequency + recency with a 1-week exponential decay
  • Ghost text inline — uses zsh's POSTDISPLAY widget, not a separate pane
  • Daemon architecture — one lightweight background process serves all terminal windows; <1ms response per keystroke
  • Local-only — all data stays in a local SQLite database; nothing leaves your machine
  • Alternatives picker — press Tab to cycle through ranked alternatives without leaving the line

Installation

Homebrew (macOS & Linux)

brew install Giammarco-Ferranti/deja/deja && deja import && (grep -qF 'deja init zsh' ~/.zshrc 2>/dev/null || echo 'eval "$(deja init zsh)"' >> ~/.zshrc) && exec zsh

curl (any Linux/macOS, no Homebrew required)

curl -fsSL https://raw.githubusercontent.com/Giammarco-Ferranti/deja/main/install.sh | sh

Both commands install deja, import your existing zsh history, add the integration to ~/.zshrc (idempotent), and reload your shell. To audit the curl installer before running it, view it on GitHub.

Oh My Zsh

If you manage zsh with Oh My Zsh, enable deja the idiomatic way, the same flow as zsh-autosuggestions. The binary still comes from Homebrew or the curl script; the plugin just sources deja's integration for you.

# 1. Install the deja binary. Skip (or remove) the `eval "$(deja init zsh)"` line it
#    offers to add to ~/.zshrc, since the plugin runs that for you:
brew install Giammarco-Ferranti/deja/deja          # or the curl installer above

# 2. Clone the plugin into Oh My Zsh's custom plugins dir:
git clone https://github.com/Giammarco-Ferranti/deja \
  ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/deja

# 3. Add `deja` to plugins=(...) in ~/.zshrc:
#      plugins=( ... deja )

# 4. Import your history once and reload:
deja import && exec zsh

zinit

If you use zinit, add this to your .zshrc:

zinit ice wait"0" lucid depth=1
zinit light Giammarco-Ferranti/deja

zinit handles the Oh My Zsh plugin integration, so the eval "$(deja init zsh)" line is not needed. Make sure the deja binary is installed separately via Homebrew or the curl installer.

Manual (any framework)

Prefer not to clone the whole repo? Fetch just the plugin file instead:

mkdir -p ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/deja
curl -fsSL https://raw.githubusercontent.com/Giammarco-Ferranti/deja/main/deja.plugin.zsh \
  -o ${ZSH_CUSTOM:-~/.oh-my-zsh/custom}/plugins/deja/deja.plugin.zsh

Pick one activation, not both. The plugin runs eval "$(deja init zsh)" for you, so if the installer already appended that line to ~/.zshrc, remove it. Keeping both double-sources the integration.

Deja replaces zsh-autosuggestions, so don't list both in plugins=(). If deja detects zsh-autosuggestions is loaded it stands down (see Troubleshooting).


Setup

The install commands above already do this for you. If you skipped them and have the binary on $PATH some other way, run these once to import your zsh history and activate the integration:

deja import
eval "$(deja init zsh)"

By default deja import reads your zsh history from $HISTFILE (when it's exported), falling back to ~/.zsh_history. If your history lives elsewhere or $HISTFILE is set in ~/.zshrc without export, so child processes can't see it point deja at the file explicitly:

deja import --file /path/to/history

To make it permanent, add the eval line to your ~/.zshrc:

# ~/.zshrc
eval "$(deja init zsh)"

Deja auto-spawns its daemon on first use and keeps it running across sessions.


Key Bindings

Key Action Rebind with
(right arrow) Accept full suggestion DEJA_ACCEPT_KEY (extra key)
Ctrl+→ Accept next word only DEJA_WORD_ACCEPT_KEY (extra key)
Shift+→ Cycle fuzzy preset forward (tight → smart → loose) DEJA_CYCLE_FUZZY_KEY
Shift+← Cycle fuzzy preset backward (loose → smart → tight) DEJA_CYCLE_FUZZY_BACK_KEY
Tab Open inline alternatives picker DEJA_CYCLE_KEY
Ctrl+X Suppress current suggestion (session-wide) DEJA_TOGGLE_KEY
(unbound) Dismiss the current ghost (this line only) DEJA_DISMISS_KEY

Accept the ghost suggestion with , not Enter. Enter executes whatever's literally in your buffer; is what commits the ghost into the buffer first.

Custom key bindings

Every binding above can be remapped by exporting the matching env var before eval "$(deja init zsh)" in your ~/.zshrc. Values are zle key sequences (e.g. ^I is Tab, ^X is Ctrl+X, ^[[1;2C is Shift+→); run bindkey -L or pipe a keypress through cat -v to discover a key's sequence. Set any var to empty to leave that key unbound.

# defaults
export DEJA_CYCLE_KEY='^I'   # Tab    → cycle alternatives
export DEJA_TOGGLE_KEY='^X'  # Ctrl+X → suppress for the session
# these are unbound by default (→ and Ctrl+→ already accept via wrapped widgets):
export DEJA_ACCEPT_KEY=       # accept full suggestion on a dedicated key
export DEJA_WORD_ACCEPT_KEY=  # accept the next word on a dedicated key
export DEJA_DISMISS_KEY=      # clear the ghost for this line (unlike Ctrl+X, the session keeps suggesting)

DEJA_DISMISS_KEY (deja-clear) differs from DEJA_TOGGLE_KEY (deja-toggle): dismiss only wipes the ghost on the current line, while toggle suppresses suggestions for the whole session until you toggle back or start a new shell.

Examples:

# Use Tab to accept the suggestion (and free Tab from cycling):
export DEJA_ACCEPT_KEY='^I' DEJA_CYCLE_KEY=

# Move alternatives-cycling off Tab so fzf/native completion keeps it:
export DEJA_CYCLE_KEY='^N'

Esc as dismiss: binding DEJA_DISMISS_KEY='^[' (Esc) directly is discouraged — Esc is the prefix byte for arrow keys, function keys, and vi-mode, so a bare ^[ binding can break those. Prefer a non-prefix key (e.g. ^G), or set it knowing the tradeoff.


Tuning fuzziness

Deja's matcher accepts any in-order subsequence of the characters you've typed. By default it stops the typed letters from sprawling too far apart in a candidate — gco will match git checkout, but won't match git remote add origin. Pick a preset to tune that strictness:

Preset Behavior Example: typing gco
loose typed letters can be far apart (up to 8 chars between) gcogit checkout -- README
smart typed letters stay close together (up to 4 chars between) — default gcogit checkout main
tight typed letters must be near-adjacent (up to 1 char between) gcogco, g.co, gc.o

Change the preset on the fly (takes effect immediately, persists across restarts):

deja fuzzy           # show current preset + examples
deja fuzzy tight     # set the preset
deja fuzzy cycle     # advance to the next preset  (tight → smart → loose → tight)
deja fuzzy back      # step to the previous preset (loose → smart → tight → loose)

Or cycle without leaving the line — press Shift+→ (forward) or Shift+← (backward) at any prompt and the next preset is applied immediately. The ghost suggestion repaints under the new mode in the same frame, and a picker-style confirmation appears below the prompt showing where you are in the ladder:

deja: fuzzy    tight    *smart*    loose

Rebind via DEJA_CYCLE_FUZZY_KEY / DEJA_CYCLE_FUZZY_BACK_KEY (set either to empty to disable that direction; tmux users may need set -g xterm-keys on for the default Shift+arrow sequences to pass through).

Or override per shell session via environment variable:

export DEJA_FUZZY=smart   # before the daemon starts; takes precedence over the saved preset

Troubleshooting

Every subcommand supports --help (e.g. deja query --help) for flag-level details. The most common issues:

Suggestions aren't appearing.

  1. Check the daemon is reachable: deja ping should print pong.
  2. Confirm the integration is loaded in your shell: eval "$(deja init zsh)" must be in ~/.zshrc and the shell re-sourced (exec zsh).
  3. Ctrl+X toggles per-session suppression — start a new shell to clear it.

Using another inline-suggestion plugin. Deja renders its own ghost text and replaces zsh-autosuggestions — don't run both. If Deja detects zsh-autosuggestions is loaded it prints a one-line notice and stands down (rather than wrapping the same ZLE widgets, which can wedge the line editor). To use Deja, remove zsh-autosuggestions from plugins=() in ~/.zshrc and restart your shell.

The daemon seems stuck.

A fresh terminal will auto-respawn it via the init script.

Stale socket after a crash.

rm ~/.local/share/deja/sock

Then open a new shell.

Reset the database (start over from current ~/.zsh_history).

pkill -f 'deja daemon'
rm ~/.local/share/deja/deja.db
deja import

Where data lives.

Path Purpose
~/.local/share/deja/deja.db SQLite database (history, stats, sequences)
~/.local/share/deja/sock Unix socket the daemon listens on
~/.local/share/deja/init.zsh Generated zsh integration script

How It Works

Deja is built around four signals that are combined into a single composite score:

score = 1.0 × fuzzy
      + 0.4 × frecency
      + 0.3 × directory_affinity
      + 0.5 × sequence_score
Signal What it measures
Fuzzy Subsequence match quality with bonuses for consecutive characters, word boundaries, and prefix hits
Frecency Log-scaled frequency combined with exponential recency decay (1-week half-life)
Directory affinity How often you've run this command from the current directory
Sequence score Probability that this command follows the one you just ran

Architecture

┌─────────────────┐     JSON/Unix socket      ┌──────────────────────┐
│   zsh widget    │ ──────────────────────▶   │   deja daemon        │
│  (per keystroke)│ ◀──────────────────────   │  (single process,    │
└─────────────────┘    suggestion (<1ms)       │   all terminals)     │
                                               └──────────┬───────────┘
                                                          │
                                                    SQLite (WAL)
                                               commands · stats · seqs

The daemon loads all state into memory at startup (map[string]*CommandStat, top-100 directory affinities, sequence pairs) and uses a sync.RWMutex so reads never block each other. Writes (command recording) take microseconds.

If the daemon is unavailable, deja query falls back to a direct SQLite read automatically.


Building from Source

git clone https://github.com/Giammarco-Ferranti/deja.git
cd deja
make build        # produces ./bin/deja

go test ./...     # run all tests
go vet ./...      # lint

Releases

Releases are automated via release-please and driven by conventional commits on main:

  • feat: ... → minor bump
  • fix: ... → patch bump
  • feat!: ... or a BREAKING CHANGE: footer → major bump
  • chore:, docs:, test:, refactor: → no version bump

After qualifying commits land on main, the release-please workflow opens (and keeps updating) a Release PR that bumps .release-please-manifest.json and updates CHANGELOG.md. Merging that PR is the release action — it creates the vX.Y.Z git tag, which triggers release.yml to run the test suite and (only on green) publish binaries via GoReleaser and update the Homebrew tap.

Maintainers should not run git tag manually.


Contributing

Contributions are welcome — see CONTRIBUTING.md for setup, workflow, and commit conventions. For anything larger than a small fix, please open an issue first so we can align on direction.

The scorer (internal/scorer/) is the most iteration-heavy part of the codebase — the four signal weights are the best place to experiment if you want to improve suggestion quality.

Security

Please report vulnerabilities privately via GitHub's "Report a vulnerability" button on the repo's Security tab, not as public issues.


Uninstall

  1. Remove the integration line from ~/.zshrc:
  2. Stop the running daemon:
  3. Delete local data (history DB, socket, generated init script):
    rm -rf ~/.local/share/deja/
  4. Remove the binary, depending on how you installed it:
    • Homebrew: brew uninstall deja (and optionally brew untap Giammarco-Ferranti/deja)
    • curl installer: rm "$(which deja)" (default location is ~/.local/bin/deja)

License

MIT — see LICENSE.


Made with ☕ and a friendly ghost.