The official dkod plugin for Claude Code.
Documentation • Quickstart • Multi-Agent Workflows • Discord
Two commands. Everything you need in one plugin.
/plugin marketplace add dkod-io/dkod-plugin
/plugin install dkod@dkod
On first use, a browser window opens for GitHub OAuth. After that, dkod tools are available in all sessions.
Not using Claude Code? Install the dkod skill instead — works with Cursor, Windsurf, Cline, Codex, and any MCP-compatible agent.
| Component | What you get |
|---|---|
| MCP Server | 12 tools — dk_connect, dk_context, dk_file_read, dk_file_write, dk_file_list, dk_submit, dk_verify, dk_approve, dk_merge, dk_push, dk_status, dk_watch |
| Skill | Teaches agents to decompose work by symbol, launch concurrent sub-agents, handle conflicts |
| Agent | parallel-executor — orchestrates multi-agent workflows with automatic landing |
| Commands | /dkod:status /dkod:push /dkod:watch /dkod:land |
| Hooks | SubagentStart — reminds sub-agents to create isolated dkod sessions |
AI agents serialize work to avoid conflicts. If two tasks touch the same file, one waits for the other. Git sees code as text — two edits to the same file means a merge conflict, even if the changes are completely independent.
Your agents are fast. Git is holding them back.
The dkod plugin teaches your agent a new default: parallelize everything.
|
No clones, no forks, no branches per agent. dkod gives each agent an isolated session overlay — a lightweight, copy-on-write layer on top of a single shared codebase. One repo. Unlimited agents. Zero overhead. |
dkod merges at the code structure level, not the text level. Two agents editing different functions in the same file? Automatic merge in under 50ms. True conflicts are caught and surfaced with full semantic context. |
| Scenario | Result |
|---|---|
| Two agents edit different functions in the same file | Auto-merge |
| Two agents add different fields to the same struct | Auto-merge |
| Two agents add the same import | Deduplicated |
| Two agents modify different sections of a function | Auto-merge |
| Two agents modify the same function body | Conflict (surfaced with context) |
| Agent A deletes a function that Agent B calls | Conflict (caught at merge time) |
- Connect — each agent gets an isolated session (
dk_connect) - Work — agents read, write, and query code independently
- Submit — agents submit their changes (
dk_submit) - Land — approve, merge, and push in one step (
/dkod:land)
Or do it manually: dk_approve → dk_merge → dk_push
Two agents editing different functions in the same file? Auto-merged. Same import added twice? Deduplicated. True semantic conflict? Surfaced with full context — never silently overwritten.
Agents receive live notifications via dk_watch when other agents modify the same file. Warnings are tagged [AFFECTS YOUR WORK] with specific symbol names — so agents can adapt in real time instead of discovering conflicts at merge.
dkod-plugin/
├── .claude-plugin/
│ ├── plugin.json # Plugin manifest (v1.1.0)
│ └── marketplace.json # Marketplace catalog
├── .mcp.json # MCP server → api.dkod.io/mcp
├── skills/
│ └── dkod/
│ ├── SKILL.md # Parallel execution behavioral guide
│ └── references/
│ └── mcp-workflow.md # Full MCP protocol reference
├── agents/
│ └── parallel-executor.md # Multi-agent orchestrator
├── commands/
│ ├── status.md # /dkod:status
│ ├── push.md # /dkod:push
│ ├── watch.md # /dkod:watch
│ └── land.md # /dkod:land (auto-pipeline)
└── hooks/
└── hooks.json # SubagentStart session reminder
MIT — free to use, fork, and build on.
Built for the age of agent-native development • dkod.io