CLI Flag Reference

This is the complete top-level flag set for wayland-core, verified against the engine source. Run wayland-core --help to see it in your installed version.

wayland-core [OPTIONS] [PROMPT]...

The trailing [PROMPT]... is the task. With a prompt, the engine runs single-shot; without one, it opens the interactive session.

Provider and model

Flag	Short	Env	Description
--provider	-p	PROVIDER	Provider to use, for example anthropic or openai.
--api-key	-k	API_KEY	API key for the provider.
--base-url	-b	BASE_URL	Base URL for the API (used by OpenAI-compatible providers).
--model	-m	MODEL	Model name.
--max-tokens			Max output tokens per response.
--max-turns			Max agent loop turns.
--system-prompt			Custom system prompt for the run.
--profile			Named profile from the config file.

Agents

Flag	Description
--agent <NAME>	Adopt a built-in agent persona (for example architect, debugger), loading its system prompt and turn limit unless overridden.
--list-agents	List the built-in agent personas and exit.

Permissions

Flag	Description
--auto-approve	Approve all tool executions, skipping the confirmation prompt.
--force	Approve every tool call for the whole run with no interactive gate. Aliases: --yolo, --dangerously-skip-permissions. Use only for trusted, scripted runs.

Sessions

Flag	Short	Description
--resume <ID>		Resume a specific saved session.
--continue	-c	Resume the most recent session. Mutually exclusive with --resume and --session-id.
--session-id <ID>		Use a specific session ID instead of auto-generating one.
--list-sessions		List saved sessions and exit.

Output and mode

Flag	Description
--no-color	Disable colored output.
--json-stream	Enable JSON streaming mode for host-client integration.
--no-tui	Use the line-based REPL instead of the terminal UI.
--compaction <LEVEL>	Output compaction level: off, safe (default), or full.
--toon	Enable TOON encoding for JSON arrays (set per session, cannot change mid-conversation).

Config and project

Flag	Description
--init-config	Generate a default config file.
--config-path	Print the config file path and exit.
--skills-path	Print the skill directory paths and exit.
--project-dir <PATH>	Directory to load .wayland-core.toml from. Defaults to the current directory.

Authentication

Flag	Description
--login	Sign in with an Anthropic account (OAuth device flow).
--logout	Remove saved OAuth credentials.

Skills lifecycle

Flag	Description
--skills-audit	Run the skills audit. Writes JSON to .wayland-core/skills-audit.json and renders Markdown to stdout.
--skills-audit-stale-days <N>	Staleness threshold in days for the audit. Default 180. Requires --skills-audit.
--skills-promote <PROCEDURE_ID>	Promote a drafted skill from staged to active.
--skills-archive <PROCEDURE_ID>	Archive a staged or active skill.

Diagnostics and evolution

Flag	Description
--doctor	Run the system-dependency doctor. Probes external binaries and environment signals, with per-distro install hints. Exit code 1 if a required check fails.
--memory-show <SESSION_ID>	Print stored memory for a session.
--online-evolution	Enable live online evolution. At session end the engine emits one evolution event and applies the Paraphrase mutator to successful trajectories. Equivalent to [observability] online_evolution = true.

Replay

Flag	Description
--replay <TRACE_PATH>	Deterministically replay a recorded trace.
--replay-diff <OTHER_TRACE_PATH>	Diff against another trace. Requires --replay.
--replay-force-version-skew	Force replay across a version mismatch. Requires --replay.

Next

• CLI Subcommand Reference
• Configuration