--- title: "Agent handoff: passing work and context from one AI agent to another · Cork" description: "An agent handoff is when one AI coding agent passes unfinished work, findings or decisions to another agent or person, so the next one need not start over." canonical: "https://corkcli.com/blog/agent-handoff" last_updated: "2026-10-03" published: "2026-10-03" author: "Kevin Vera" --- # Agent handoff: passing work and context from one AI agent to another An agent handoff is when one AI coding agent passes unfinished work, findings or decisions to another agent or person, so the next one need not start over. By Kevin Vera · October 3, 2026 · [Glossary](https://corkcli.com/blog/c/glossary) · 3 min read An **agent handoff** is the moment one AI coding agent passes work to another agent or a person: the task, what has been done, what was learned, and what is left, so the receiver can continue without starting from zero. Handoffs happen constantly in multi-agent coding: a lead agent delegates to helpers, a helper reports back, a session ends mid-task, or one developer's agent needs another developer's agent to change something it owns. ## Why it matters Agents do not share memory by default. Claude Code's documentation notes that "each Claude Code session begins with a fresh context window," and that a subagent "doesn't see your conversation history, the skills you've already invoked, or the files Claude has already read." Whatever is not written into the handoff is lost, and the receiver pays to rediscover it, or gets it wrong. A weak handoff looks like "continue the refactor". A strong one says which functions were changed, which callers still need updating, which approach was tried and rejected and why, and which tests are failing. ## How it works Common handoff channels: - **Delegation prompt.** A lead agent writes the task into the helper's prompt. Claude Code describes how "Claude composes a delegation message that summarizes the task, and the subagent works from there." - **Summary back.** The helper returns a result; Claude Code subagents return "only the summary" to the main conversation. - **Messages between agents.** Claude Code's agent teams give each agent a mailbox, and "any teammate can message any other by that name." - **Shared task lists**, where unfinished work becomes a task someone else can claim. - **Files in the repository**: commit messages, TODOs, notes in AGENTS.md or CLAUDE.md, pull request descriptions. ## Examples - An agent runs out of time halfway through migrating an API. It records the three endpoints done, the two left, and the shim it added, as a task for the next session. - A developer's agent needs a new field on a model another teammate's agent is editing. Instead of editing it directly, it messages that agent with the request. - A helper agent finds that a test is flaky for reasons unrelated to its task and hands that finding to the lead rather than fixing it. ## How teams handle it - Write handoffs as structured notes: done, learned, left, risks. - Anchor notes to the code they concern, not just to a chat thread, so the next agent finds them when it touches that code. - Prefer messaging the agent that owns the code over editing around it. - Keep a shared task list for unfinished work. Cork supports handoffs as part of coordination. Cork is the [coordination layer](https://corkcli.com/blog/coordination-layer-ai-coding-agents) for multiplayer coding agents: it keeps Claude Code, Codex and Cursor agents, on one machine or across a team's machines, from overwriting each other on one codebase, and hands each agent what the others already learned. Any agent can message the others in the repo, one agent, or whoever is working on a given function, and its `remember` and `recall` tools keep durable notes anchored to code, so the next agent to touch that code is handed the decision. ## Related terms - [Agent briefing](https://corkcli.com/blog/agent-briefing) - [Multiplayer coding agents](https://corkcli.com/blog/multiplayer-coding-agents) - Coordination layer (for AI coding agents) - [Parallel AI coding agents](https://corkcli.com/blog/parallel-ai-coding-agents) ## Sources - - - ## Related reading - [Agent briefing: the context an AI coding agent gets before it starts](https://corkcli.com/blog/agent-briefing): An agent briefing is the targeted context an AI coding agent gets before it works: signatures, callers, recent changes and notes for the code it will touch. - [How to hand work from one AI coding agent to another](https://corkcli.com/blog/hand-off-work-between-ai-coding-agents): Hand off work between AI coding agents without losing context: commit, write a handoff note, resume or fork sessions in Claude Code and Codex, and verify first. - [Agent merge conflict: when parallel AI agents' branches will not merge](https://corkcli.com/blog/agent-merge-conflict): An agent merge conflict is a git merge conflict between branches written by AI coding agents running in parallel, usually found only when the work is combined. - [How to share context between Codex and Claude Code](https://corkcli.com/blog/share-context-codex-claude-code): Make Codex and Claude Code work from the same context: one AGENTS.md, shared MCP servers, handoff files, and live awareness when both edit the same repo. [All articles](https://corkcli.com/blog.md) · [Cork in Markdown](https://corkcli.com/index.md) · [llms.txt](https://corkcli.com/llms.txt) This page in HTML: https://corkcli.com/blog/agent-handoff