kungfu-systems/kungfu

Your agents don’t hand off the work. Kungfu keeps the same Work moving across Codex, Claude, OpenCode, and your own execution surface.

View on GitHub ↗Jump to charts ↓

Summary Information

Updated 35 minutes ago
Added to GitGenius on September 15th, 2026
Created on November 15th, 2017
Open Issues & Pull Requests: 92 (+0)
GitHub issues: Enabled
Number of forks: 1,288
Total Stargazers: 4,513 (+0)
Total Subscribers: 256 (+0)

Repository Insights (GitGenius)

Median issue/PR response: 2.5 days
Mean response time: 37.9 days
90th percentile: 7.3 days
Tracked items: 62

Most active contributors

Sign in to see contributor activity.

How this project is maintained

Around half of the issues opened in the past year never receive a reply. Only 18% of issues opened in the past year have been closed. Three people close 100% of everything that gets resolved.

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Open issues: 4
New in 7 days: 0
Closed in 7 days: 0
Avg open age: 30 days
Stale 30+ days: 3
Stale 90+ days: 0

Recent activity

Opened in 7 days: 0
Closed in 7 days: 0
Comments in 7 days: 0
Events in 7 days: 0

Top labels

  • buildchain-contract-drift (29)
  • area/buildchain (26)
  • kind/internal (26)
  • source/automation (26)
  • state/blocked (26)
  • kind/bug (1)
  • source/external (1)
  • state/needs-triage (1)

Most active issues this week

No issue events were indexed in the last 7 days.

Detailed Description

Kungfu is an agent orchestration framework that keeps work persistent and continuous across multiple AI agents and execution surfaces.

The problem Kungfu solves is the friction and context loss that occurs when work transfers between different agents. Each handoff typically requires copying context, re-explaining decisions, tracking updates, and verifying what was lost. Kungfu addresses this by maintaining a durable Work object outside any individual agent session, preserving the objective, progress, evidence, next action, and completion state. This allows developers to switch between agents—Codex, Claude, OpenCode, Amp, or custom execution surfaces—without losing continuity. The tool keeps the same work moving forward regardless of which agent takes over, enabling cost optimization by using specialized agents when it matters and cheaper alternatives when it does not.

Kungfu suits teams that orchestrate multiple AI agents and need reliable handoffs between them. It is particularly valuable for projects where work spans multiple agent sessions or where cost optimization requires switching between different models. The tool maintains work continuity, verified project understanding through declared sources and decision tracking, inspectable history of what happened and why, and completion authority through independent review. Developers should expect to keep the `.kungfu/` project-local workspace for durable work and runtime state, though most contents remain local and Kungfu does not automatically stage or commit files. The framework currently requires a real agent for independent review to complete end-to-end workflows, though the built-in Mock Agent can demonstrate work creation and recovery without external credentials.

The project is in Alpha and provides deterministic recovery scenarios to test local continuity and failure handling. Installation is available for macOS and Linux through a per-user installer that does not require sudo or shell profile modification, with Windows and higher-assurance installation options documented separately. The tool includes optional TUI and GUI views for managing projects and work across agent sessions, though these are sidecars rather than requirements for basic operation.