CLI Reference
RLM Code ships one CLI entry point (rlm-code) and a rich slash-command surface inside the TUI. This page reflects the current implementation in rlm_code/main.py, rlm_code/ui/tui_app.py, and rlm_code/commands/slash_commands.py.
Entry Point
rlm-code
Launch the unified TUI (RLM, Files, Details, Shell, Research).
| Flag | Short | Description |
--verbose | -v | Enable verbose logs |
--debug | | Enable debug mode with full traceback |
--version | | Print version info and exit |
--skip-safety-check | | Skip startup directory safety checks (hidden/internal) |
Examples:
rlm-code
rlm-code -v
rlm-code --debug
rlm-code --version
Startup Safety Checks
On startup, RLM Code checks the working directory:
- Warns when run from home/personal folders.
- Blocks high-risk system directories.
- Recommends running from a dedicated project workspace.
Command Surfaces
RLM Code has two command layers in the TUI:
- TUI-native commands in
tui_app.py (layout/navigation/shell shortcuts). - Full slash handler commands in
slash_commands.py (RLM, sandbox, MCP, exports, etc.).
TUI-native commands
| Command | Description |
/help | Show in-TUI quick help |
/workflow | Show recommended RLM workflow |
/connect | Open interactive connect picker (/connect acp opens ACP picker) |
/models | Show model/provider status |
/status | Refresh status panel |
/snapshot [file] | Snapshot file/project baseline |
/diff [file] | Diff against snapshot |
/view <chat\|files\|details\|shell\|research\|next\|prev> | Switch active view (chat route opens the RLM tab) |
/layout <single\|multi> | Toggle one-screen vs multi-pane layout |
/pane <files\|details\|shell> [show\|hide\|toggle] | Pane visibility |
/focus <chat\|default> | Focus controls |
/copy | Copy last assistant response |
/shell [command] | Open shell tab or run command in shell tab |
/rml ... | Alias for /rlm ... |
/exit or /quit | Exit TUI |
Shell shortcuts:
!<cmd>: run command inline in chat.><cmd>: alternate inline shell shortcut.
Full Slash Commands
Core
| Command | Description |
/init | Initialize project config and scan workspace |
/connect [provider model [api-key] [base-url]] | Connect directly, or run with no args for interactive picker |
/models | Model/provider listing |
/status | Connection + runtime status |
/disconnect | Disconnect current model |
/help | Command help |
/intro | Intro guide |
/clear | Clear conversation |
/history | Show conversation history |
/exit | Exit |
RLM
| Command | Description |
/rlm run <task> ... | Run one RLM episode |
/rlm bench ... | Run benchmark preset/list/pack |
/rlm bench compare ... | Candidate vs baseline gate compare |
/rlm bench validate ... [--json] | CI-style gate output |
/rlm bench report ... | Export compare report (markdown/csv/json) |
/rlm import-evals pack=... | Preview external eval packs |
/rlm judge pred=... ref=... | LLM-judge predictions |
/rlm frameworks | Adapter readiness table |
/rlm viz [run_id\|latest] | Trajectory tree visualization |
/rlm status [run_id] | Run status |
/rlm abort [run_id\|all] | Cooperative cancel for active runs |
/rlm replay <run_id> | Replay stored trajectory |
/rlm doctor [env=...] [--json] | Environment diagnostics |
/rlm chat <message> ... | Persistent RLM chat sessions |
/rlm chat status [session=name] | Chat memory stats |
/rlm chat reset [session=name] | Reset chat memory |
/rlm observability | Observability sink status |
Harness
| Command | Description |
/harness tools [mcp=on\|off] | List harness tools (local + optional MCP) |
/harness doctor | Harness tool coverage report |
/harness run <task> [steps=N] [mcp=on\|off] [tools=name[,name2]] | Run tool-driven coding loop |
Sandbox
| Command | Description |
/sandbox status | Runtime health + pure RLM backend status |
/sandbox doctor | Detailed runtime diagnostics |
/sandbox use <runtime> | Switch runtime (local, docker, apple-container, modal, e2b, daytona) |
/sandbox profile <secure\|dev\|custom> | Apply superbox runtime policy preset |
/sandbox backend <exec\|monty\|docker> [ack=I_UNDERSTAND_EXEC_IS_UNSAFE] | Set Pure RLM backend |
/sandbox strict <on\|off> | Toggle pure RLM strict mode |
/sandbox output-mode <truncate\|summarize\|metadata> | Control REPL output compaction |
/sandbox apple <on\|off> | Enable/disable Apple container runtime gate |
Notes:
backend=exec requires explicit ack token and sets unsafe opt-in.profile secure applies Docker-first strict defaults for research/production.
Execution and Validation
| Command | Description |
/run [timeout=N] | Execute generated code |
/validate [file] | Validate code quality/safety patterns |
/test [file] | Run tests |
MCP
| Command | Description |
/mcp-servers | List configured servers |
/mcp-connect <server> | Connect server |
/mcp-disconnect <server> | Disconnect server |
/mcp-tools [server] | List tools |
/mcp-call <server> <tool> [args] | Call tool |
/mcp-resources [server] | List resources |
/mcp-read <server> <uri> | Read resource |
/mcp-prompts [server] | List prompts |
/mcp-prompt <server> <name> [args] | Fetch prompt |
Sessions, Export, and Project
| Command | Description |
/sessions | List sessions |
/session ... | Save/load/delete session state |
/save [file] | Save latest output/code |
/export ... | Export session/package/config/conversation |
/import ... | Import session/config |
/project ... | Project info commands |
/save-data [file] | Save generated data |
Templates and DSPy helpers
| Command | Description |
/demo | Demo workflows |
/eval | Generate/evaluate code |
/optimize, /optimize-start, /optimize-status, /optimize-cancel, /optimize-resume | GEPA optimization flow |
/examples | Template examples |
/predictors, /adapters, /retrievers | DSPy component references |
/async, /streaming, /data, /explain, /reference | DSPy/usage help |
Environment Variables (Common)
| Variable | Purpose |
OPENAI_API_KEY | OpenAI model auth |
ANTHROPIC_API_KEY | Anthropic model auth |
GEMINI_API_KEY / GOOGLE_API_KEY | Gemini model auth |
TUI behavior tunables include:
RLM_TUI_HISTORY_ITEMSRLM_TUI_HISTORY_ITEM_CHARSRLM_TUI_HISTORY_TOTAL_CHARSRLM_TUI_THINK_TICKRLM_TUI_EVENT_FLUSH_SECONDSRLM_TUI_EVENT_BATCH_LIMITRLM_TUI_ACP_DISCOVERY_TIMEOUT_SECONDSRLM_TUI_ACP_CACHE_TTL_SECONDSRLM_TUI_HARNESS_AUTORLM_TUI_HARNESS_AUTO_MCPRLM_TUI_HARNESS_AUTO_STEPSRLM_TUI_HARNESS_PREVIEW_STEPSRLM_TUI_INPUT_DEBOUNCE_SECONDSRLM_TUI_CHAT_MAX_LINESRLM_TUI_TOOL_MAX_LINESRLM_TUI_EVENT_MAX_LINES
See Configuration for full config schema.