Highlight 1 / 3
Review file changes
Select code in a prepared fictional change, queue comments across two files, and send one feedback batch to pi. File references and selected code stay attached to the conversation.
Open source desktop app for pi
A desktop client
for the pi coding agent.
Run the pi coding agent with local or hosted models. Manage sessions, Git worktrees, code review, and web previews in one desktop application.
pi
AG-UI
CopilotKit
Angular
NgRx
Electron
llama.cpp
llama-swap
TailscaleProduct demos
Three workflows, start to finish.
Three short workflows. Choose one, then press play.
Created with pi-kit · Fictional projects and data · English narration: Qwen3-TTS, Aiden · English captions in the player.
Why pi-kit
A desktop interface
for pi and local models.
We love pi for its composability and model freedom. We needed a Windows desktop client for working with local models, and wanted an experience that felt more at home on Windows than a terminal interface. That practical need became pi-kit, which now runs on Windows, macOS, and Linux.
pi-kit adds model setup, credential storage, session navigation, and review tools while using pi's existing session and extension formats.
We also maintain the Angular port of CopilotKit. pi-kit uses it with AG-UI to render tool interfaces, input requests, and MCP Apps in the conversation.
Read the project storyFeatures
Models, sessions,
and development tools.
Configure models, navigate conversation history, isolate changes in worktrees, and inspect results before merging.
01Model freedom
Hugging Face models and llama-swap
Search Hugging Face and download supported GGUF models in pi-kit. Verified files join your llama-swap config, and pi-kit can start, stop, check, and repair llama-swap for you—or connect a server you already run.
Read the guide02Skill library
Find and install skills
Search the skills.sh catalog, preview instructions and bundled files, then install a GitHub skill at the reviewed revision. Choose user or trusted-project scope.
Read the guide03Code review
Git-backed change tracking
Compare thread snapshots or the working tree. See added and removed line totals in the thread header, then open the changes pill to review file diffs.
Read the guide04Parallel work
Durable background agents
Let the main agent hand side tasks—tests, reviews, research—to background agents that keep running across restarts. Follow them in the Agents panel and answer their questions when they need you.
Read the guide05Beyond the editor
Computer use
Opt in to let pi see the screen and operate native apps on macOS, Windows, and Linux. Looking runs freely; acting asks first.
Read the guide06Session history
Session branching
Navigate pi’s session tree, branch from an earlier turn, and keep the original approach available while you explore another.
Read the guide07Isolated work
Git worktrees
Give a thread its own Git branch and folder. Inspect the diff, sync with the base, and merge back when the change is ready.
Read the guide08Interactive tools
A2UI and MCP Apps
Use A2UI and sandboxed MCP Apps for interactive tool interfaces, with explicit prompts when the agent needs your decision.
Read the guide09Secrets
OS credential storage
API keys and connector tokens go to Keychain, Credential Manager, or Secret Service. Config files hold only a resolver command.
Read the guide10Reusable knowledge
Skill drafting and verification
Draft a skill from a finished thread, replay it in a scratch worktree, and review the result before approving what gets installed.
Read the guide11Live preview
Artifacts browser
Open your running project in Artifacts. You and the agent navigate, inspect, and test the same native page.
Read the guide12Visual feedback
Screenshots and inspection
Attach a page or element screenshot, DOM details, or selected console logs to your unsent draft.
Read the guideArtifacts
An integrated browser
for web development.
Open a running web project beside the conversation. The agent can navigate, interact with, and inspect the same browser page displayed in Artifacts.
Explore the Artifacts browserStart your project server, enter its HTTP address, and navigate the same page the agent can operate. Resize the dock to check narrower layouts.
Browser tools
Screenshots, inspection,
and console logs.
Attach a viewport, full-page, or selected-element screenshot to a message. Include DOM details or selected console entries when needed. Attachments are added to the unsent draft.
Screenshots, inspection, and logsChoose the scope that explains the issue. Use selected-element capture for a small detail or include surrounding layout with a visible-area image.
01 / SEE
Capture the actual state
Use the page you have open, at the width where the problem happens. No need to describe a layout from memory.
02 / EXPLAIN
Add precise context
Attach element details or selected console messages when the issue needs more than pixels. Review private information before sending.
03 / VERIFY
Repeat the interaction
Ask the agent to revisit the same flow, check the console, and capture the result. A passing build is not visual verification.
Background agents
Side tasks that keep
running in the background.
The main agent hands tests, reviews, and research to background agents. They start read-only by default and can ask you or the main agent for a decision. Follow their progress in the Agents panel, review their reports, and stop runs when needed. Runs use the selected model and may incur provider costs.
Work with background agentsEach card shows the agent’s role, permission mode, running time, turns, tokens, and spend. Open, stop, or delete it from the card.
Remote access
Use pi-kit
from your phone.
Publish pi-kit on your own Tailscale network and pair your phone with a code you approve on the computer. Nothing is exposed to the internet and no traffic passes through a pi-kit server. The computer has to stay awake with pi-kit open.
Set up remote accessInstall Tailscale on both devices, approve HTTPS certificates once, choose a port, and enable remote access. pi-kit only reads Tailscale’s status until you do.
Local models
Download GGUF models
from Hugging Face.
Search supported public GGUF models on Hugging Face from pi-kit. Review the source, license, file size, and memory estimate, then confirm the download; pi-kit verifies its SHA-256 checksum. Verified downloads are added to your selected llama-swap config.
Existing local servers can be connected without downloading another model. Hosted connections include Radius, supported subscriptions, provider API keys, and LiteLLM.
- Start, stop, and repair llama-swap from Settings
- Pinned revisions and SHA-256 verification
- Hardware-aware estimates and existing llama-swap integration
Code review
Review changes.
Compare file diffs.
Thread changes are tracked with Git snapshots. Compare a thread snapshot or the working tree, and open review from the added/removed line count in the thread header. Move between files, compare unified or split diffs, and search the selected diff. The screenshots below show the earlier 0.8.3 reviewer.
Explore code reviewChoose the change source and file, step backward or forward through the file list, and keep the review controls above the diff.
Themes and local setup
Appearance options.
Local engine setup.
Choose Soverius or CopilotKit in Appearance. Find platform-specific engine installation help, check the llama.cpp runtimes your models use, and inspect grouped local provider connections.
Explore SettingsA dark blue palette and matching fonts. Change the theme in Settings → Appearance → Look.
Sessions and Git
Session branches
and code review.
Navigate earlier turns and alternate conversation branches. Use worktrees to isolate file changes. Session history and Git history are separate.
Worktrees and code reviewReturn to earlier turns and inspect another approach without losing the original conversation. Navigating history does not undo filesystem changes.
Self-learning
Reusable skills
from completed tasks.
Self-learning in pi-kit creates instructions for future tasks, not new model weights. The bundled @pi-kit/learn package analyzes a thread and drafts a standard pi SKILL.md file. You review the instructions before installing them.
- 01
Record the result
A completed thread has the conversation, tool calls, and checks used to make a change. A recorded commit, merge, or pull-request outcome provides the base and result for verification.
- 02
Draft the procedure
Run
/learn, or let the enabled attendant draft from a qualifying outcome. The analysis uses a redacted transcript and outcome diff to extract steps, encountered errors, and verification commands into a proposed skill. - 03
Replay and compare
Optionally run the draft in a scratch worktree at the outcome's base commit, using the first prompt. The report compares changed files and hunks and reruns recognized project checks. Reproduction can pass, be partial, or fail.
- 04
Edit and approve
Read the proposed
SKILL.md, edit the procedure, and select project or user scope. Approve installs it; Discard leaves it uninstalled. An automatically generated draft is not automatically approved. - 05
Use it in another thread
Select the approved skill from the composer or
/menu and pin it when needed. Later messages receive those instructions as context. Settings → Skills → Library keeps its origin, usage, and verification record.
The task changed the starter function and added three tests. Read opens the generated skill.
Self-learning example project
A dependency-free Node.js project with a starter function and three tests. Complete a normalization task, draft and verify a skill, then select it for a second task. The walkthrough includes the prompts, Git steps, and results to inspect.
Drafting and replay use configured models and may incur provider costs. A replay does not merge or push changes; its tools are not sandboxed to the scratch checkout. A passing comparison applies to that replay, not every future use.
Security
Credential storage
and configuration.
API keys and connector tokens go to macOS Keychain, Windows Credential Manager, or Linux Secret Service. pi's auth.json and mcp.json only hold a command that resolves the secret when needed.
OS credential store Encrypted database
~/.pi/agent/auth.json
{
"openrouter": {
"type": "api_key",
"key": "!pi-kit-secret get openrouter"
}
} No API key is stored in this file
Extensions
Packages, extensions,
and skills.
Browse and install pi packages from npm, Git, or a local folder inside Settings. See the extensions, skills, prompts, and themes each package brings before you add it.
- Enable, update, or remove in place
- User and project scopes
- Open sessions reload automatically
Download pi-kit
Get the latest release from GitHub. Choose the asset for your platform: Windows x64 or ARM64, macOS Apple Silicon, or Linux x64 or ARM64.
Release notes, installer files, and available SHA-256 digests are listed on GitHub.
On a server: npm install -g @soverius-ai/pi-kit, or a server archive from the same release. See Install on a server.
From the team behind pi-kit
Bring AI into
your team’s work.
We build AI applications and offer team workshops in AI engineering.
Learn together
Hands-on team workshops
Build skills in AI engineering, agentic applications, and software development with AI.
Explore workshops Build togetherEngineering for your AI project
Work with us on architecture, agentic interfaces, backend systems, and local or hybrid models.
Explore our services