AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
Code Assistant/
PI-Desktop
PI-Desktop logo

PI-Desktop

Active·★ 8.0k·LGPL-3.0·Updated 2026-11-17

PI-Desktop is a local-first desktop workspace for coding-agent projects, sessions, models and plugins. It combines Electron with a Rust host core and the pi agent harness, offering an independent interface rather than requiring a specific IDE. The product is relevant when persistent conversations, project review and replaceable providers matter more than staying inside a terminal. We examined a fixed November 5 commit whose root package reports 0.21.0; the README labels this line Early Preview. The repository license is LGPL-3.0.

Eleven local checks exercised the actual filesystem and preference modules using synthetic profiles and Node.js 26.6.0 on macOS. Overlapping destinations, a symbolic-link target and unrelated nonempty directories were rejected. Session files copied exactly while originals remained readable. Browser-root installation locks were omitted, while user lockfiles in other locations survived. Selected internal and external symlinks retained their intended targets after remapping. A source changed at verification was rejected; an injected relocation failure propagated without removing the original session. Valid preferences round-tripped and malformed state was refused.

These are file-layer results. The fixture maps Electron original-fs to Node physical filesystem calls and replaces Rust relocation with a recording or failing callback. It does not open a database, start Electron, restore an agent session or test Windows junctions. File copying therefore cannot establish full application migration. Try a disposable project, verify attachments and reopen the app using the new storage before discarding backups. Plugins and model connections need separate evaluation. Local-first describes workspace state, while cloud model requests still go to the configured provider; no offline, model-quality or plugin-isolation result is implied.

Read the full review, verification scope and sources →

Hands-on tested · Tested on 2026-11-05 with 0.21.0; commit 836ccd084147bedf916e56c859c1529ae63d2ac6 (filesystem and preference modules only).

Setup: macOS arm64, Node.js 26.6.0; synthetic temporary profiles, original-fs mapped to node:fs, Rust relocation callback substituted

Commands run

node doc/review-fixtures/mobile-desktop-20261105/verify.mjs --sources <source-parent> --output /tmp/mobile-desktop-results.json

What we saw

  • Eleven selected checks passed: target validation, exact session copy, original preservation and root-only installation-lock exclusions.
  • Selected symlinks retained targets; source mutation rejected, injected relocation failure propagated; malformed preferences refused.
  • No Electron startup, real database rewriting, Windows junctions, crash recovery, full build, model or plugin execution.

Data as of 2026-11-12.

License
LGPL-3.0
Latest release
v0.22.0 · 2026-11-08
Last commit
2026-11-11
Contributors
133
Forks
772
Open issues and PRs
239
First published
2023-03-22

Source: GitHub, npm and PyPI public APIs.

#agent-framework#ai-agent#coding-agent#desktop-app#electron#global#i18n#local-first
↗ Visit site★ GitHub
01

Features

01Keeps project sessions across app launches with search, branching and diff review.
02Offers Agent, Plan and Goal modes for different task scopes.
03Extends tools, panels, MCP servers and background services through plugins.
04Connects cloud models, local models and custom gateways.
02

Why choose it

+Selected file migration preserved original sessions and rejected overlapping or unrelated targets.
+Scoped lock exclusions retained user files; tested symlinks kept their selected targets.
+Source-change detection and explicit preference validation made selected failures visible.
03

Trade-offs

!The 0.21.x line is Early Preview; a full Electron/Rust migration was not tested.
!File migration requires stopped writers and can leave partial destination files after failure.
!External symlinks can keep dependencies outside the migrated root; plugin isolation remains untested.
05

Maintenance check

ActiveChecked: 2026-11-05
Last commit
2026-11-05
Commits, last 90 days
100+
Releases, last 90 days
30
Latest release
v0.22.0-beta.5 (2026-11-04)
Open issues
267
Recent issues with at least one comment
8/23
Commits per week, last 12 weeks
Commits per week, last 12 weeks: 140, 157, 204, 632, 529, 415, 256, 206, 190, 119, 155, 98

Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

  • #2000 招募:用 AI 对 PI-Desktop 插件文档做 review 与补齐(PR 必须关联本 issue)18 comments · 2026-10-30
  • #247 Thank you for using PI-Desktop — we read every report16 comments · 2026-09-11
  • #1645 [Bug] 远程服务器无法使用13 comments · 2026-10-14
  • #1948 [Bug] 对话缓存异常9 comments · 2026-10-28
  • #1530 [Bug] Win11 25H2版本下,升级到beta版出现无法安装7 comments · 2026-10-10

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

04

Who it fits

Pick it if
  • You want an independent desktop workspace with persistent projects and replaceable model providers.
  • You can trial preview releases and verify a complete storage move and restart on your platform.
Look elsewhere if
  • You require a validated production storage recovery path from this selected-module review.
  • You expect local-first to mean that cloud-model requests never leave the device.
06

Use cases

↳Read code, edit files, run commands and iterate on tests in a desktop workspace.
↳Inspect an implementation plan before running a refactor.
↳Split a larger task across inspectable worker sessions with separate contexts.
07

Alternatives

⚡Top VS:
VSATLAS★2.1k→VSdsh-Go★26.0k→VSKun★6.4k→VScommand-code★4.2k→
ATLAS logo
ATLAS★ 2.1k
Adaptive Test-time Learning and Autonomous Specialization
VSCompare→
dsh-Go logo
dsh-Go★ 26.0k
dsh-go — DeepSeek Harness 装进口袋:手机本机运行,也能遥控你的电脑。
VSCompare→
Kun logo
Kun★ 6.4k
Local-first AI agent workspace for coding, writing, design, research, and automation — one runtime for desktop GUI and TUI.
VSCompare→
command-code logo
command-code★ 4.2k
Command Code AI — the best coding agent for open models.
VSCompare→
CoreCoder logo
CoreCoder★ 1.9k
Minimal AI coding agent (~1,000 lines of Python) inspired by Claude Code. Works with any LLM. Think NanoGPT for coding agents. Formerly NanoCoder.
zero logo
zero★ 1.8k
The coding agent that answers to you, your model, your machine, your rules.
VSCompare→
opencode-telegram-bot logo
opencode-telegram-bot★ 1.4k
OpenCode mobile client via Telegram: run and monitor AI coding tasks from your phone while everything runs locally on your machine. OpenCode V2 support.
VSCompare→
easy-stock logo
easy-stock★ 1.9k
A股行情分析与AI智能投研智能体:股票分析、量化交易分析、盘后复盘桌面工作台——easy stock
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs05Maintenance check04Who it fits06Use cases07Alternatives
vastsa /
PI-Desktop
↗

A persistent desktop agent workspace with explicit storage migration

8.0k772LGPL-3.0
Last commit1d ago
StatusActive
CategoryCode Assistant
Contributors133
Latest releasev0.22.0
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.