Core Concepts

Harnesses

A harness is the coding agent the CLI installs into. Each component type lands in the place that harness already reads from at runtime — the CLI never imposes its own layout. The CLI auto-detects the harness in the current folder; override it with --harness <id>.

Supported harnesses

Ten harnesses are supported. The first three take every component type the CLI syncs; the other seven take skills, instructions and files only.

Harness	Component types
`claude-code` (Claude Code)	skills, mcps, agents, commands, playbooks, instructions, files
`codex` (Codex)	skills, mcps, commands, playbooks, instructions, files
`kafka` (Kafka)	skills, mcps, agents, commands, playbooks, instructions, files
`qwen-code` (Qwen Code)	skills, instructions, files
`opencode` (OpenCode)	skills, instructions, files
`openclaw` (OpenClaw)	skills, instructions, files
`cursor` (Cursor)	skills, instructions, files
`factory` (Factory Droid)	skills (inlined into `AGENTS.md`), instructions, files
`qoder` (Qoder)	skills, instructions, files
`pi` (Pi)	skills, instructions, files

For the seven newer harnesses the CLI writes no MCP config, commands, sub-agents or playbooks; a component of those types is skipped when it is installed into one of them.

Where each component lands

Component	Claude Code	Codex	Kafka
Skill	`.claude/skills/<slug>/`	`~/.codex/skills/<slug>/` or `.agents/skills/<slug>/`	`.kafka/skills/<slug>/`
MCP server	`.mcp.json` (project) / `~/.claude.json` → `mcpServers` (global)	`~/.codex/config.toml` `[mcp_servers.<slug>]`	`.kafka/kafka.json`
Command	`.claude/commands/<slug>.md`	`~/.codex/prompts/<slug>.md`	`.kafka/commands/<slug>.md`
Sub-agent	`.claude/agents/<slug>.md`	—	`.kafka/agents/<slug>.md`
Playbook	`.claude/playbooks/<slug>.md` + table in `CLAUDE.md`	`.codex/playbooks/<slug>.md` + table in `AGENTS.md`	`.kafka/playbooks/<slug>.md` + table in `KAFKA.md`
Instruction	Block in `CLAUDE.md`	Block in `AGENTS.md`	Block in `KAFKA.md`
File	Arbitrary copy to a target path	Same	Same

The newer harnesses, project scope first and global scope second. Each global root can be moved with the environment variable in the last column.

Harness	Skill	Instructions	Global root
`qwen-code`	`.qwen/skills/<slug>/` · `~/.qwen/skills/<slug>/`	Block in `QWEN.md` · `~/.qwen/QWEN.md`	`QWEN_HOME`
`opencode`	`.opencode/skills/<slug>/` · `~/.opencode/skills/<slug>/`	Block in `AGENTS.md` · `~/.opencode/AGENTS.md`	`OPENCODE_HOME`
`openclaw`	`skills/<slug>/` at the folder root · `~/.openclaw/skills/<slug>/`	Block in `AGENTS.md` · `~/.openclaw/AGENTS.md`	`OPENCLAW_HOME`
`cursor`	`.cursor/rules/<slug>.mdc` · `~/.cursor/rules/<slug>.mdc`	Block in `AGENTS.md` · `~/.cursor/AGENTS.md`	`CURSOR_HOME`
`factory`	Inlined as a block in the instructions file	Block in `AGENTS.md` · `~/.factory/AGENTS.md`	`FACTORY_HOME`
`qoder`	`.qoder/skills/<slug>/` · `~/.qoder/skills/<slug>/`	Block in `AGENTS.md` · `~/.qoder/AGENTS.md`	`QODER_HOME`
`pi`	`.pi/skills/<slug>/` · `~/.pi/agent/skills/<slug>/`	Block in `AGENTS.md` · `~/.pi/agent/AGENTS.md`	`PI_CODING_AGENT_DIR`

Marker-based instruction blocks

Instruction and playbook blocks are bracketed by  / :end markers in the harness's instructions file, and the playbook table is regenerated on every pull — so removed playbooks drop out automatically and your hand-written prose is never clobbered.

Force a harness or install scope for onboard / sync with --harness <id> and --scope <global|project>. Conversation tracking is available for claude-code and codex only.