feat(cli): REPL commands for permissions/sessions/hooks/trust + --dangerously-skip-permissions

CHANGELOG.md

@@ -30,6 +30,10 @@ cost estimates, and a much more robust terminal UI.
   OpenAI/DeepSeek/Kimi/Gemini automatic caches keep hitting.
 - **`/cost`** and cost estimates in the per-turn line, `/tokens`, the bottom bar and JSON output; prices are
   approximate and overridable under `pricing:` in `config.yaml`.
+- **More slash commands**, matching the CLI: `/permissions allow|deny|set|reset`, `/sessions`, `/hooks`, `/trust`,
+  `/untrust`, `/exit`.
+- **`--mode` for `devorch` / `devorch chat`, and `--dangerously-skip-permissions`** on every entry point (same as
+  `--mode auto`: no prompts; dangerous-command blocks and deny rules still apply).
 - **Input surface**: open input box with a coloured mode label, `Shift+Tab` mode cycling, `Ctrl+J` newline, a
   width-aware placeholder and bottom bar, context-usage gauge.
 
@@ -42,6 +46,7 @@ cost estimates, and a much more robust terminal UI.
 - The bottom toolbar re-ran `git` on every keystroke; it is now cached.
 - The bottom toolbar rendered as a bright reversed bar, and was clipped on narrow terminals, hiding the mode.
 - Saving config no longer drops a `pricing:` section.
+- The slash-command dropdown vanished when a command was fully typed (`/cost`, `/models`).
 
 ### Changed
 - CI runs on pull requests only.

README.md

@@ -447,7 +447,23 @@ With `--output json` stdout contains a single JSON object (`result`, `is_error`,
 on failure, so it is safe to use in scripts and CI. `stream-json` emits `tool_call` events as they happen
 followed by a final `result` event. Available on `ask` and `run`.
 
-All non-interactive commands support `--provider`, `--model`, `--mode`, `--mcp`, and `--no-mcp`.
+All non-interactive commands support `--provider`, `--model`, `--mode`, `--mcp`, `--no-mcp` and
+`--dangerously-skip-permissions`. `devorch` / `devorch chat` take `--mode` and `--dangerously-skip-permissions`
+too, so you can start the REPL in a given mode.
+
+### Skipping permission prompts
+
+```bash
+devorch --dangerously-skip-permissions        # start the REPL in AUTO mode
+devorch ask --dangerously-skip-permissions "run the tests and fix failures"
+devorch --mode plan                           # or start in plan / ask mode
+```
+
+`--dangerously-skip-permissions` is the same as `--mode auto`: **no prompts at all**, so the agent can edit files
+and run commands unasked. It is not a full bypass: the built-in dangerous-command blocks (`rm -rf *`, `sudo`, ...)
+and your own `deny` rules still apply, and project hooks still run. It conflicts with `--mode ask|plan`, and it
+overrides a project's `default_mode`. In the REPL, `/auto` and `/ask` switch the same behaviour on and off, and
+`/permissions allow --session shell pytest*` skips prompts for just one kind of command.
 
 ## All Slash Commands
 
@@ -484,7 +500,12 @@ All non-interactive commands support `--provider`, `--model`, `--mode`, `--mcp`,
 | `/mcp start <name>` | Reconnect a server from config |
 | `/mcp stop <name>` | Disconnect a server and remove its tools |
 | `/config` | Show configuration settings |
-| `/permissions` | Show and manage tool permission settings |
+| `/permissions` | Show permissions; change them with `/permissions allow\|deny [--session] <tool> <pattern>`, `set <tool> <allow\|deny\|ask>`, `reset` |
+| `/sessions` | `list [n]`, `show <id>`, `delete <id>` (resume with `devorch --resume <id>`) |
+| `/hooks` | Show active hooks and whether project hooks are trusted |
+| `/trust` / `/untrust` | Trust (or stop trusting) this project's `.devorch` settings and hooks |
+| `/cost` | Estimated session cost per model |
+| `/exit` | Exit DevOrch (also `exit`, `quit`, or Ctrl+C twice) |
 | `/tasks` | Show multi-step task list and execution progress |
 
 ## Configuration

cli/commands/_shared.py

@@ -303,8 +303,27 @@ def print_project_notices(settings) -> None:
         print_warning(notice)
 
 
-def resolve_mode(mode_str: str | None, default: AgentMode = AgentMode.AUTO) -> AgentMode:
-    """Parse and validate a mode string, exit with error on bad input."""
+SKIP_PERMISSIONS_HELP = (
+    "Skip every permission prompt (same as --mode auto). Built-in dangerous-command blocks and "
+    "your deny rules still apply."
+)
+
+
+def resolve_mode(
+    mode_str: str | None,
+    default: AgentMode = AgentMode.AUTO,
+    skip_permissions: bool = False,
+) -> AgentMode:
+    """Parse and validate a mode string, exit with error on bad input.
+
+    ``skip_permissions`` (``--dangerously-skip-permissions``) forces AUTO, which never prompts; deny
+    rules and the built-in dangerous-command blocks are still enforced by the executor.
+    """
+    if skip_permissions:
+        if mode_str and mode_str.lower() != "auto":
+            print_error(f"--dangerously-skip-permissions conflicts with --mode {mode_str}")
+            raise typer.Exit(1)
+        return AgentMode.AUTO
     if not mode_str:
         project_mode = load_project_mode()
         return AgentMode(project_mode) if project_mode else default

cli/commands/ask.py

@@ -4,6 +4,7 @@
 
 from cli.commands._output import configure_output, run_with_output, validate_format
 from cli.commands._shared import (
+    SKIP_PERMISSIONS_HELP,
     build_agent,
     build_tools,
     console,
@@ -30,6 +31,9 @@ def ask(
     mode: str = typer.Option(None, "--mode", help="Execution mode: ask, auto, plan"),
     no_mcp: bool = typer.Option(False, "--no-mcp", help="Disable all MCP servers for this run"),
     mcp: list[str] = typer.Option(None, "--mcp", help="Use only these MCP servers (repeatable)"),
+    skip_permissions: bool = typer.Option(
+        False, "--dangerously-skip-permissions", help=SKIP_PERMISSIONS_HELP
+    ),
     output: str = typer.Option(
         "text", "--output", "-o", help="Output format: text, json, stream-json"
     ),
@@ -56,7 +60,7 @@ def ask(
 
     llm = create_provider(provider, model, settings)
     tools = build_tools(settings, no_mcp=no_mcp, mcp_only=mcp or None)
-    agent_mode = resolve_mode(mode, default=AgentMode.AUTO)
+    agent_mode = resolve_mode(mode, default=AgentMode.AUTO, skip_permissions=skip_permissions)
     memory_ctx = MemoryManager().get_context_prompt()
     project_ctx = ProjectContextLoader().load()
     project_prompt = project_ctx.to_system_prompt_block() if project_ctx else ""

cli/commands/edit.py

@@ -3,6 +3,7 @@
 import typer
 
 from cli.commands._shared import (
+    SKIP_PERMISSIONS_HELP,
     build_agent,
     build_tools,
     console,
@@ -23,6 +24,9 @@ def edit(
     mode: str = typer.Option("auto", "--mode", help="Execution mode: ask, auto, plan"),
     no_mcp: bool = typer.Option(False, "--no-mcp", help="Disable all MCP servers for this run"),
     mcp: list[str] = typer.Option(None, "--mcp", help="Use only these MCP servers (repeatable)"),
+    skip_permissions: bool = typer.Option(
+        False, "--dangerously-skip-permissions", help=SKIP_PERMISSIONS_HELP
+    ),
 ):
     """
     Edit a specific file using an AI instruction.
@@ -42,7 +46,7 @@ def edit(
     llm = create_provider(provider, model, settings)
     # edit only needs file-oriented tools — no terminal/task tools
     tools = build_tools(settings, include_terminal=False, no_mcp=no_mcp, mcp_only=mcp or None)
-    agent_mode = resolve_mode(mode)
+    agent_mode = resolve_mode(mode, skip_permissions=skip_permissions)
     memory_ctx = MemoryManager().get_context_prompt()
     agent = build_agent(llm, tools, memory_ctx, agent_mode, pricing=settings.pricing)
 

cli/commands/run.py

@@ -4,6 +4,7 @@
 
 from cli.commands._output import configure_output, run_with_output, validate_format
 from cli.commands._shared import (
+    SKIP_PERMISSIONS_HELP,
     build_agent,
     build_tools,
     console,
@@ -28,6 +29,9 @@ def run(
     ),
     no_mcp: bool = typer.Option(False, "--no-mcp", help="Disable all MCP servers for this run"),
     mcp: list[str] = typer.Option(None, "--mcp", help="Use only these MCP servers (repeatable)"),
+    skip_permissions: bool = typer.Option(
+        False, "--dangerously-skip-permissions", help=SKIP_PERMISSIONS_HELP
+    ),
     output: str = typer.Option(
         "text", "--output", "-o", help="Output format: text, json, stream-json"
     ),
@@ -62,7 +66,7 @@ def run(
     final_prompt = render_prompt(skill_data, extra)
 
     tools = build_tools(settings, no_mcp=no_mcp, mcp_only=mcp or None)
-    agent_mode = resolve_mode(mode)
+    agent_mode = resolve_mode(mode, skip_permissions=skip_permissions)
     memory_ctx = MemoryManager().get_context_prompt()
     project_ctx = ProjectContextLoader().load()
     project_prompt = project_ctx.to_system_prompt_block() if project_ctx else ""

cli/constants.py

@@ -44,7 +44,12 @@
     "/clear": "Clear conversation history (saves accomplishments to project memory)",
     "/session": "Show current session ID, model, and message count",
     "/config": "Show configuration settings and provider status",
-    "/permissions": "Show and manage tool permission settings",
+    "/permissions": "Show or change permissions (/permissions allow|deny|set|reset)",
+    "/sessions": "List, show or delete chat sessions (/sessions list|show <id>|delete <id>)",
+    "/hooks": "Show active hooks and whether project hooks are trusted",
+    "/trust": "Trust this project's .devorch settings and hooks (MCP, allow rules, hooks)",
+    "/untrust": "Revoke trust for this project's .devorch files",
+    "/exit": "Exit DevOrch (also: exit, quit, Ctrl+C twice)",
     "/compact": "Summarize and compact conversation history to save tokens",
     "/models": "Search, browse, and switch models (interactive 15/page pagination)",
     "/model": "Switch model directly (/model <name>) or open interactive search",