{"slug": "show-hn-kungfu-keep-coding-agent-work-alive-across-sessions-and-handoffs", "title": "Show HN: Kungfu – Keep coding-agent work alive across sessions and handoffs", "summary": "Kungfu, an alpha-stage tool from Kungfu Systems, lets coding agents such as Codex, Claude, OpenCode, and Amp share durable work across sessions and handoffs, preserving objectives, progress, and evidence. The tool, installed via a per-user script without sudo, includes a deterministic Mock Agent scenario to test recovery from disconnects and crashes, and supports both staying in a current agent or launching one through Kungfu.", "body_md": "**Your agents don't hand off the work. You do.**\n\nEvery switch means copying context, re-explaining decisions, chasing updates, and checking what got lost. Kungfu keeps the same work moving, no matter which agent takes over.\n\nUse the best Agent when it matters. Use a cheaper one when it does not. Keep the same Work across Codex, Claude, OpenCode, Amp, or your own execution surface.\n\nKungfu UNGFU™· Never Guess. Facts Unfold.[Why this signature exists].\n\nKungfu is currently Alpha. On macOS or Linux, install Kungfu and make the `kungfu`\n\ncommand available with the reviewed per-user installer:\n\n```\ncurl -fsSL https://kungfu.tech/install.sh | sh\n```\n\nIt does not use `sudo`\n\nor edit your shell profile. Follow the exact `PATH`\n\nstep\nit prints, then open a project and run the deterministic recovery story:\n\n```\ncd your-project\nKUNGFU_MOCK_AGENT_SCENARIO=recovery-story kungfu\n```\n\nThe built-in Mock Agent needs no provider credentials. It starts the same Work through three consecutive Attempts: a disconnect, a crash, then a recovered delivery. Because the scenario is deterministic, it tests Kungfu's local continuity and recovery path rather than a particular model.\n\nFor a quicker onboarding check that crosses a question, an approval, and a ready-for-review result in one Attempt, run:\n\n```\nKUNGFU_MOCK_AGENT_SCENARIO=multi-step kungfu\n```\n\nSee the [installation guide](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/installing-cli.md) for Windows,\nhigher-assurance installation, explicit version pinning, and troubleshooting.\n\nThe Mock Agent covers Work creation and execution. Regular onboarding still needs a supported real Agent for independent review, so this is not yet an end-to-end zero-external-Agent path.\n\n**Stay in your current Agent.** Paste this sentence into Codex, Claude, OpenCode,\nAmp, or another Agent that can run local commands:\n\n```\nRun `kungfu agent brief`, then guide me through my first Project and Work. Keep me in my current agent, and use Kungfu as the durable Work layer.\n```\n\n**Start an Agent through Kungfu.** Use its familiar native console while Kungfu\nkeeps the Work behind it:\n\n```\ncd your-project\nkungfu run codex\n```\n\nUse `claude`\n\n, `opencode`\n\n, or `amp`\n\ninstead of `codex`\n\nwhen that is the agent you\nalready use. Pass a task to create the first Work directly:\n\n```\nkungfu run codex \"Prepare the release notes\"\n```\n\n**When .kungfu/ appears.** It is Kungfu's project-local workspace for durable\nWork and runtime state. Do not delete it or add the whole directory to Git. Ask\nyour Agent to run\n\n`kungfu agent map --json`\n\nand follow its `workspaceGit`\n\npolicy\nbefore staging anything. Most contents stay local; Kungfu never stages,\ncommits, or pushes files for you.**Open the optional global view later.** The Kungfu TUI and GUI can show and\nmanage Projects and Work across Agent Sessions. They are sidecar views, not a\nrequirement for every Agent conversation.\n\nKungfu keeps the Work—not the chat—outside every Agent Session: its objective, progress, evidence, next action, and completion state.\n\n**Work continuity.** Change the Agent without losing what was done, what remains, or which Work is continuing.**Verified project understanding.** Declared sources, important omissions, conflicts, decisions, and uncertainty remain visible instead of being guessed.**Inspectable history.** People and Agents can see what happened, what the next action is, and where the supporting evidence came from.**Completion authority.** An Agent can produce a result and its evidence; independent review and Kungfu settlement decide whether Work is complete.\n\nThe first visible value should arrive within minutes. The deeper value appears over the following days, when the same Work survives context loss, changed understanding, failed attempts, handoff, and restart.\n\nA **Project** remembers where related Work belongs. A **Work** keeps one durable\nobjective and its current truth. An **Attempt** records what one Agent tried,\nincluding failure, without replacing or erasing the Work.\n\nThe same Work state is available from native Agent consoles, the Kungfu TUI and GUI, the CLI, and APIs. If another live Agent already owns a Work, Kungfu stops a second writer instead of letting two Agents silently diverge.\n\nDurable Work must answer three questions in order: can it survive a new Agent, can it survive failure, and who is allowed to complete it?\n\nYes. The first proof isolates continuity: one Work continues across two fresh Agent Sessions without copied chat.\n\nYes. The second proof moves from mechanism to real failure conditions. Inside a disposable Project, the connection drops, a new process resumes, that process crashes, and both Attempts remain under the same Work.\n\nWork survival is only the first step. If an Agent can declare its own result complete, continuity still is not trustworthy.\n\nOnly the governed review and settlement path—not the Agent that did the work. The third proof separates Agent exit, independent review, and Kungfu settlement. An Agent can produce the candidate and evidence; it cannot approve its own Work.\n\nThese are bounded exact-artifact demonstrations—not provider rankings,\nproduction certification, or authority to complete real Work. See the\n[animation technical specification and auditable evidence](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/qualification/auditable-demo-artifact-pipeline.md).\n\nTo enter the Lab yourself instead of watching an artifact, run\n`kungfu agent-work-lab`\n\n. Its short `open → watch/tour → try → test → report`\n\njourney is documented in the [Agent Work Lab guide](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/agent-work-lab.md).\n\nOpen the optional terminal view whenever you want the larger Work picture:\n\n```\nkungfu\n```\n\nGetting Started leads to the same Agent-first prompt, so this remains a sidecar rather than a replacement for your familiar Agent interface.\n\nYou do not need to learn every Kungfu subsystem before making a bounded change. From a source checkout, give your Agent the task you actually want to complete:\n\n```\nRead `AGENTS.md`. I want to <task>. Use the repository's verified task-context route. Before editing, explain only the concepts, current implementation owners, authority boundaries, and qualification path this task requires.\n```\n\nIf you are exploring or evaluating the repository rather than changing it, paste this into your Agent:\n\n```\nInspect https://github.com/kungfu-systems/kungfu.\nRead AGENTS.md first. Explain what Kungfu does and evaluate its architecture.\nGive me only the smallest mental model I need; do not make me learn the\nrepository's full ontology.\n```\n\nFor a whole-system explanation, start with the [Evolution\nMap](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/evolution/README.md). For a bounded change, Shifu compiles a verified\n[Agent Task Chart](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/xinfa-agent-context.md) and expands it only when\nthe task requires more context. Required omissions, stale authority, and\nambiguous routing remain visible instead of being filled by guesswork.\n\nThe README stops at the product entry. Detailed implementation, evidence, and claim boundaries live in their own maintained routes:\n\n**Use and understand Kungfu:**[System Overview](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/concepts/system-overview.md), the[Documentation Guide](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/README.md), and the complete[Documentation Map](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/MAP.md).**Build or contribute:**[Contributing to Kungfu](/kungfu-systems/kungfu/blob/dev/v4/v4.0/CONTRIBUTING.md),[AGENTS.md](/kungfu-systems/kungfu/blob/dev/v4/v4.0/AGENTS.md),[Verified Context for Agents](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/xinfa-agent-context.md), and[runtime surface provenance](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/concepts/runtime-surface-provenance.md).**Verify guarantees and current limits:**[Contracts](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/qualification/contracts.md),[Known Limits](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/qualification/known-limits.md), and the[KFD support matrix](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/qualification/kfd-support-matrix.md).**Evaluate release, update, and migration behavior:**[Upgrade Kungfu](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/upgrading.md)and the[Exit and version compatibility policy](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/exit-and-version-compatibility.md).**Evaluate the wider ecosystem thesis:** the[Agent Supply Chain architecture](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/architecture/agent-supply-chain.md).\n\nKungfu v4 is publicly available in Alpha. Download the\n[current release artifacts](https://github.com/kungfu-systems/kungfu/releases/latest)\nor follow the [installation guide](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/installing-cli.md). Alpha remains\na prerelease channel, not a stable or generally available release; exact\nsupport, qualification, and non-claims remain in\n[Alpha Status](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/guides/alpha-status.md) and\n[Known Limits](/kungfu-systems/kungfu/blob/dev/v4/v4.0/docs/qualification/known-limits.md).\n\n- Product home:\n[https://kungfu.tech](https://kungfu.tech) - Developer and agent surface:\n[https://libkungfu.dev](https://libkungfu.dev) - Issues and questions:\n[GitHub issue forms](https://github.com/kungfu-systems/kungfu/issues/new/choose) - Security reports:\n[SECURITY.md](/kungfu-systems/kungfu/blob/dev/v4/v4.0/SECURITY.md) - License:\n[Apache License 2.0](/kungfu-systems/kungfu/blob/dev/v4/v4.0/LICENSE)\n\nKungfu's protocol, release infrastructure, build environments, and public sites are developed in the open:\n\n[Buildchain](https://github.com/kungfu-systems/buildchain)— auditable build and release infrastructure with verifiable Release Passports.[KFD](https://github.com/kungfu-systems/kfd)— the open engineering standard for reliable action and continuity under uncertainty.[Build Images](https://github.com/kungfu-systems/build-images)— source for the reproducible environments used to build Kungfu system artifacts.[kungfu.tech source](https://github.com/kungfu-systems/site-kungfu-tech)— source for the public product site.[libkungfu.dev source](https://github.com/kungfu-systems/site-libkungfu-dev)— source for the developer and agent surface.[All public repositories](https://github.com/kungfu-systems)— the complete organization-level source map.", "url": "https://wpnews.pro/news/show-hn-kungfu-keep-coding-agent-work-alive-across-sessions-and-handoffs", "canonical_source": "https://github.com/kungfu-systems/kungfu", "published_at": "2026-08-16 13:53:42+00:00", "updated_at": "2026-08-16 14:11:02.872696+00:00", "lang": "en", "topics": ["developer-tools", "ai-agents", "ai-tools"], "entities": ["Kungfu", "Kungfu Systems", "Codex", "Claude", "OpenCode", "Amp"], "alternates": {"html": "https://wpnews.pro/news/show-hn-kungfu-keep-coding-agent-work-alive-across-sessions-and-handoffs", "markdown": "https://wpnews.pro/news/show-hn-kungfu-keep-coding-agent-work-alive-across-sessions-and-handoffs.md", "text": "https://wpnews.pro/news/show-hn-kungfu-keep-coding-agent-work-alive-across-sessions-and-handoffs.txt", "jsonld": "https://wpnews.pro/news/show-hn-kungfu-keep-coding-agent-work-alive-across-sessions-and-handoffs.jsonld"}}