Agent Handoff is a GitHub-native standard for passing project context between AI coding agents, human maintainers, and human-supervised agents.
AI coding agents often lose context between sessions, tools, branches, and pull requests. Chat history is not enough for medium repositories where several humans and agents may work in parallel.
Agent Handoff keeps durable context where development already happens:
GitHub = Issues, Pull Requests, reviews, checks, labels, comments, ownership
Git = branches, commits, diffs, tags, history
ai/ = compact durable project memory and agent protocols
flowchart LR
I[Issue] --> C[Work Claim]
C --> B[Branch]
B --> P[Draft PR]
P --> T[Checks]
T --> H[Handoff]
H --> N[Next agent or human]
- maintainers coordinating AI-assisted development;
- developers using Codex-like coding agents;
- teams working with ChatGPT, Cursor, Claude Code, local agents, or custom LLM tools;
- projects that need visible ownership, compact memory, and safe handoff between runs.
Start with the English documentation: docs/en/README.md.
Add Agent Handoff to this repository.
Use the latest standard from https://github.com/artyomboyko/Agent_Handoff.
Inspect the current repository first.
Create or update the Agent Handoff files, GitHub Issue Forms, and Pull Request template.
Keep the repository language English unless the downstream project explicitly chooses another language.
Open a Pull Request and leave a compact handoff.
- GitHub is the source of work truth.
- Git is the source of code truth.
ai/is compact durable memory.- Handoffs are short, structured, and reviewable.
- Humans stay in control.
| Area | File |
|---|---|
| Standard | AGENT_HANDOFF_STANDARD.md |
| GitHub workflow | ai/GITHUB_WORKFLOW.md |
| Agent guide | AGENTS.md |
| Memory map | ai/README.md |
| Work claim | ai/WORK_CLAIM_PROTOCOL.md |
| Task reports | ai/TASK_REPORT_PROTOCOL.md |
| Agent identity | ai/AGENT_IDENTITY.md |
| Refactoring workflow | ai/REFACTORING.md |
| Issue labels | ISSUE_LABELS.md |
| Issue status | ISSUE_STATUS.md |
| FAQ | FAQ.md |
| Examples | examples/ |
| Approach | What it keeps | Limitation |
|---|---|---|
| Chat memory | Conversation context | Tool-specific and session-bound |
| README only | Project overview | Not enough for active work ownership |
| Long LOG.md | Detailed history | Becomes noisy and hard to review |
| Wiki | Documentation | Often detached from branches and PRs |
| Agent Handoff | Compact project state, claims, handoffs | Requires small workflow discipline |
Agent Handoff is related to AI coding agents, Codex-like agents, ChatGPT coding workflows, Cursor, Claude Code, LLM agents, project context, agent memory, GitHub workflow, multi-agent development, handoff protocol, pull request workflow, and human-agent collaboration.
Use Agent Handoff to see who owns work, what changed, what was tested, what remains risky, and where the next contributor or agent should continue.
Start from AGENTS.md, read the required files, claim work in GitHub, open a Draft PR early, keep ai/ compact, and leave a handoff when work is completed, paused, blocked, or transferred.
This repository is public. Before reusing the standard in your own public repository, review repository history, generated files, and workflow logs.
License: GPL-3.0. See CHANGELOG.md for version history.