@oh-my-pi/pi-coding-agent 3.37.0 → 4.0.0

package/CHANGELOG.md

@@ -2,6 +2,95 @@
 
 ## [Unreleased]
 
+## [4.0.0] - 2026-01-10
+
+### Added
+
+- Exported `InteractiveModeOptions` type for programmatic SDK usage
+- Exported additional UI components for extensions: `ArminComponent`, `AssistantMessageComponent`, `BashExecutionComponent`, `BranchSummaryMessageComponent`, `CompactionSummaryMessageComponent`, `CustomEditor`, `CustomMessageComponent`, `FooterComponent`, `ExtensionEditorComponent`, `ExtensionInputComponent`, `ExtensionSelectorComponent`, `LoginDialogComponent`, `ModelSelectorComponent`, `OAuthSelectorComponent`, `SessionSelectorComponent`, `SettingsSelectorComponent`, `ShowImagesSelectorComponent`, `ThemeSelectorComponent`, `ThinkingSelectorComponent`, `ToolExecutionComponent`, `TreeSelectorComponent`, `UserMessageComponent`, `UserMessageSelectorComponent`
+- Exported `renderDiff`, `truncateToVisualLines`, and related types for extension use
+- `setFooter()` and `setHeader()` methods on `ExtensionUIContext` for custom footer/header components
+- `setEditorComponent()` method on `ExtensionUIContext` for custom editor components
+- `supportsUsageInStreaming` model config option to control `stream_options: { include_usage: true }` behavior
+- Terminal setup documentation for Kitty keyboard protocol configuration (Ghostty, wezterm, Windows Terminal)
+- Documentation for paid Cloud Code Assist subscriptions via `GOOGLE_CLOUD_PROJECT` env var
+- Environment variables reference section in README
+- `--no-tools` flag to disable all built-in tools, enabling extension-only setups
+- `--no-extensions` flag to disable extension discovery while still allowing explicit `-e` paths
+- `blockImages` setting to prevent images from being sent to LLM providers
+- `thinkingBudgets` setting to customize token budgets per thinking level
+- `PI_SKIP_VERSION_CHECK` environment variable to disable new version notifications at startup
+- Anthropic OAuth support via `/login` to authenticate with Claude Pro/Max subscription
+- OpenCode Zen provider support via `OPENCODE_API_KEY` env var and `opencode/<model-id>` syntax
+- Session picker (`pi -r`) and `--session` flag support searching/resuming by session ID (UUID prefix)
+- Session ID forwarding to LLM providers for session-based caching (used by OpenAI Codex for prompt caching)
+- `dequeue` keybinding (`Alt+Up`) to restore queued steering/follow-up messages back into the editor
+- Pluggable operations for built-in tools enabling remote execution via SSH or other transports (`ReadOperations`, `WriteOperations`, `EditOperations`, `BashOperations`, `LsOperations`, `GrepOperations`, `FindOperations`)
+- `/model <search>` pre-filters the model selector or auto-selects on exact match; use `provider/model` syntax to disambiguate
+- Managed binaries directory (`~/.omp/bin/`) for fd and rg tools
+- `FooterDataProvider` for custom footers with `getGitBranch()`, `getExtensionStatuses()`, and `onBranchChange()`
+- `ctx.ui.custom()` accepts `{ overlay: true }` option for floating modal components
+- `ctx.ui.getAllThemes()`, `ctx.ui.getTheme(name)`, `ctx.ui.setTheme(name | Theme)` for theme management
+- `setActiveTools()` for dynamic tool management
+- `setModel()`, `getThinkingLevel()`, `setThinkingLevel()` methods for runtime model and thinking level changes
+- `ctx.shutdown()` for requesting graceful shutdown
+- `pi.sendUserMessage()` for sending user messages from extensions
+- Extension UI dialogs (`select`, `confirm`, `input`) support `timeout` option with live countdown display
+- Extension UI dialogs accept optional `AbortSignal` to programmatically dismiss dialogs
+- Async extension factories for dynamic imports and lazy-loaded dependencies
+- `user_bash` event for intercepting user `!`/`!!` commands
+- Built-in renderers used automatically for tool overrides without custom `renderCall`/`renderResult`
+- `InteractiveMode`, `runPrintMode()`, `runRpcMode()` exported for building custom run modes
+- Copy link button on messages for deep linking to specific entries
+- Codex injection info display showing system prompt modifications
+- URL parameter support for `leafId` and `targetId` deep linking
+- Wayland clipboard support for `/copy` command using wl-copy with xclip/xsel fallback
+
+### Changed
+
+- Bash tool output truncation now recalculates on terminal resize instead of using cached width
+- Web search tool headers updated to match Claude Code client format for better compatibility
+- `discoverSkills()` return type documented as `{ skills: Skill[], warnings: SkillWarning[] }` in SDK docs
+- Default model for OpenCode provider changed from `claude-sonnet-4-5` to `claude-opus-4-5`
+- Terminal color mode detection defaults to truecolor for modern terminals instead of 256color
+- System prompt restructured with XML tags and clearer instructions format
+- `before_agent_start` event receives `systemPrompt` in the event object and returns `systemPrompt` (full replacement) instead of `systemPromptAppend`
+- `discoverSkills()` returns `{ skills: Skill[], warnings: SkillWarning[] }` instead of `Skill[]`
+- `ctx.ui.custom()` factory signature changed from `(tui, theme, done)` to `(tui, theme, keybindings, done)`
+- `ExtensionRunner.initialize()` signature changed from options object to positional params `(actions, contextActions, commandContextActions?, uiContext?)`
+
+### Fixed
+
+- Wayland clipboard copy (`wl-copy`) no longer blocks when the process doesn't exit promptly
+- Empty `--tools` flag now correctly enables all built-in tools instead of disabling them
+- Bash tool handles spawn errors gracefully instead of crashing the agent
+- Components properly rebuild their content on theme change via `invalidate()` override
+- `setTheme()` triggers a full rerender so previously rendered components update with new theme colors
+- Session ID updates correctly when branching sessions
+- External edits to `settings.json` while pi is running are preserved when pi saves settings
+- Default thinking level from settings applies correctly when `enabledModels` is configured
+- LM Studio compatibility for OpenAI Responses tool strict mapping
+- Symlinked directories in `prompts/` folders are followed when loading prompt templates
+- String `systemPrompt` in `createAgentSession()` works as a full replacement instead of having context files and skills appended
+- Update notification for bun binary installs shows release download URL instead of npm command
+- ESC key works during "Working..." state after auto-retry
+- Abort messages show correct retry attempt count
+- Antigravity provider returning 429 errors despite available quota
+- Malformed thinking text in Gemini/Antigravity responses where thinking content appeared as regular text
+- `--no-skills` flag correctly prevents skills from loading in interactive mode
+- Overflow-based compaction skips if error came from a different model or was already handled
+- OpenAI Codex context window reduced from 400k to 272k tokens to match Codex CLI defaults
+- Context overflow detection recognizes `context_length_exceeded` errors
+- Key presses no longer dropped when input is batched over SSH
+- Clipboard image support works on Alpine Linux and other musl-based distros
+- Queued steering/follow-up messages no longer wipe unsent editor input
+- OAuth token refresh failure no longer crashes app at startup
+- Status bar shows correct git branch when running in a git worktree
+- Ctrl+V clipboard image paste works on Wayland sessions
+- Extension directories in `settings.json` respect `package.json` manifests
+
+## [3.37.1] - 2026-01-10
+
 ## [3.37.0] - 2026-01-10
 ### Changed
 

package/README.md

@@ -9,6 +9,7 @@ Works on Linux, macOS, and Windows (requires bash; see [Windows Setup](#windows-
 - [Getting Started](#getting-started)
   - [Installation](#installation)
   - [Windows Setup](#windows-setup)
+  - [Terminal Setup](#terminal-setup)
   - [API Keys & OAuth](#api-keys--oauth)
   - [Quick Start](#quick-start)
 - [Usage](#usage)
@@ -106,6 +107,30 @@ For most users, [Git for Windows](https://git-scm.com/download/win) is sufficien
 }
 ```
 
+### Terminal Setup
+
+Pi uses the [Kitty keyboard protocol](https://sw.kovidgoyal.net/kitty/keyboard-protocol/) for reliable modifier key detection. Most modern terminals support this protocol, but some require configuration.
+
+**Kitty, iTerm2:** Work out of the box.
+
+**Ghostty:** Add to your Ghostty config (`~/.config/ghostty/config`):
+
+```
+keybind = alt+backspace=text:\x1b\x7f
+keybind = shift+enter=text:\n
+```
+
+**wezterm:** Create `~/.wezterm.lua`:
+
+```lua
+local wezterm = require 'wezterm'
+local config = wezterm.config_builder()
+config.enable_kitty_keyboard = true
+return config
+```
+
+**Windows Terminal:** Does not support the Kitty keyboard protocol. Shift+Enter cannot be distinguished from Enter. Use Ctrl+Enter for multi-line input instead. All other keybindings work correctly.
+
 ### API Keys & OAuth
 
 **Option 1: Auth file** (recommended)
@@ -169,6 +194,7 @@ omp
 - Gemini CLI uses the production Cloud Code Assist endpoint (standard Gemini models)
 - Antigravity uses a sandbox endpoint with access to Gemini 3, Claude (sonnet/opus thinking), and GPT-OSS models
 - Both are free with any Google account, subject to rate limits
+- Paid Cloud Code Assist subscriptions: set `GOOGLE_CLOUD_PROJECT` or `GOOGLE_CLOUD_PROJECT_ID` env var to your project ID
 
 Credentials stored in `~/.omp/agent/auth.json`. Use `/logout` to clear.
 
@@ -196,7 +222,7 @@ The agent reads, writes, and edits files, and executes commands via bash.
 | Command                   | Description                                                                 |
 | ------------------------- | --------------------------------------------------------------------------- |
 | `/settings`               | Open settings menu (thinking, theme, queue mode, toggles)                   |
-| `/model`                  | Switch models mid-session (fuzzy search, arrow keys, Enter to select)       |
+| `/model`                  | Switch models mid-session. Use `/model <search>` or `provider/model` to prefilter/disambiguate. |
 | `/export [file\|--copy]`  | Export session to HTML file or copy to clipboard                            |
 | `/share`                  | Upload session as secret GitHub gist, get shareable URL (requires `gh` CLI) |
 | `/session`                | Show session info: path, message counts, token usage, cost                  |
@@ -239,7 +265,7 @@ The agent reads, writes, and edits files, and executes commands via bash.
 | Key                       | Action                       |
 | ------------------------- | ---------------------------- |
 | Enter                     | Send message                 |
-| Shift+Enter / Alt+Enter   | New line (Ctrl+Enter on WSL) |
+| Shift+Enter               | New line (Ctrl+Enter on Windows Terminal) |
 | Ctrl+W / Option+Backspace | Delete word backwards        |
 | Ctrl+U                    | Delete to start of line      |
 | Ctrl+K                    | Delete to end of line        |
@@ -320,8 +346,11 @@ omp -r              # Short form
 omp --no-session    # Ephemeral mode (don't save)
 
 omp --session /path/to/file.jsonl  # Use specific session file
+omp --session a8ec1c2a             # Resume by session ID (partial UUID)
 ```
 
+**Resuming by session ID:** The `--session` flag accepts a session UUID (or prefix). Session IDs are visible in filenames under `~/.omp/agent/sessions/<project>/` (e.g., `2025-12-13T17-47-46-817Z_a8ec1c2a-5a5f-4699-88cb-03e7d3cb9292.jsonl`). The UUID is the part after the underscore. You can also search by session ID in the `omp -r` picker.
+
 ### Context Compaction
 
 Long sessions can exhaust context windows. Compaction summarizes older messages while keeping recent ones.
@@ -493,6 +522,7 @@ Add custom models (Ollama, vLLM, LM Studio, etc.) via `~/.omp/agent/models.json`
 | `supportsStore`           | Whether provider supports `store` field     |
 | `supportsDeveloperRole`   | Use `developer` vs `system` role            |
 | `supportsReasoningEffort` | Support for `reasoning_effort` parameter    |
+| `supportsUsageInStreaming` | Whether provider supports `stream_options: { include_usage: true }`. Default: `true` |
 | `maxTokensField`          | Use `max_completion_tokens` or `max_tokens` |
 
 **Live reload:** The file reloads each time you open `/model`. Edit during session; no restart needed.
@@ -831,9 +861,11 @@ omp [options] [@files...] [messages...]
 | `--continue`, `-c`                    | Continue most recent session                                                                                                                                                 |
 | `--resume`, `-r`                      | Select session to resume                                                                                                                                                     |
 | `--models <patterns>`                 | Comma-separated patterns for Ctrl+P cycling. Supports glob patterns (e.g., `anthropic/*`, `*sonnet*:high`) and fuzzy matching (e.g., `sonnet,haiku:low`)                     |
+| `--no-tools`                          | Disable all built-in tools                                                                                                                                                   |
 | `--tools <tools>`                     | Comma-separated tool list (default: `read,bash,edit,write`)                                                                                                                  |
 | `--thinking <level>`                  | Thinking level: `off`, `minimal`, `low`, `medium`, `high`                                                                                                                    |
-| `--hook <path>`                       | Load a hook file (can be used multiple times)                                                                                                                                |
+| `--extension <path>`, `-e`            | Load an extension file (can be used multiple times)                                                                                                                          |
+| `--no-extensions`                     | Disable extension discovery (explicit `-e` paths still work)                                                                                                                 |
 | `--no-skills`                         | Disable skills discovery and loading                                                                                                                                         |
 | `--skills <patterns>`                 | Comma-separated glob patterns to filter skills (e.g., `git-*,docker`)                                                                                                        |
 | `--export <file> [output]`            | Export session to HTML                                                                                                                                                       |
@@ -892,6 +924,15 @@ omp --tools read,grep,find,ls -p "Review the architecture"
 omp --export session.jsonl output.html
 ```
 
+### Environment Variables
+
+| Variable | Description |
+|----------|-------------|
+| `ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, etc. | API keys for providers (see [API Keys & OAuth](#api-keys--oauth)) |
+| `PI_CODING_AGENT_DIR` | Override the agent config directory (default: `~/.omp/agent`) |
+| `PI_SKIP_VERSION_CHECK` | Skip new version check at startup (useful for Nix or other package manager installs) |
+| `VISUAL`, `EDITOR` | External editor for Ctrl+G (e.g., `vim`, `code --wait`) |
+
 ---
 
 ## Tools

package/docs/extensions.md

@@ -260,7 +260,7 @@ pi starts
       ▼
 user sends prompt ─────────────────────────────────────────┐
   │                                                        │
-  ├─► before_agent_start (can inject message, append to system prompt)
+  ├─► before_agent_start (can inject message, modify system prompt)
   ├─► agent_start                                          │
   │                                                        │
   │   ┌─── turn (repeats while LLM calls tools) ───┐       │
@@ -311,6 +311,8 @@ pi.on("session_start", async (_event, ctx) => {
 });
 ```
 
+**Examples:** [claude-rules.ts](../examples/extensions/claude-rules.ts), [custom-header.ts](../examples/extensions/custom-header.ts), [file-trigger.ts](../examples/extensions/file-trigger.ts), [status-line.ts](../examples/extensions/status-line.ts), [todo.ts](../examples/extensions/todo.ts), [tools.ts](../examples/extensions/tools.ts)
+
 #### session_before_switch / session_switch
 
 Fired when starting a new session (`/new`) or switching sessions (`/resume`).
@@ -332,6 +334,8 @@ pi.on("session_switch", async (event, ctx) => {
 });
 ```
 
+**Examples:** [confirm-destructive.ts](../examples/extensions/confirm-destructive.ts), [dirty-repo-guard.ts](../examples/extensions/dirty-repo-guard.ts), [status-line.ts](../examples/extensions/status-line.ts), [todo.ts](../examples/extensions/todo.ts)
+
 #### session_before_branch / session_branch
 
 Fired when branching via `/branch`.
@@ -349,6 +353,8 @@ pi.on("session_branch", async (event, ctx) => {
 });
 ```
 
+**Examples:** [confirm-destructive.ts](../examples/extensions/confirm-destructive.ts), [dirty-repo-guard.ts](../examples/extensions/dirty-repo-guard.ts), [git-checkpoint.ts](../examples/extensions/git-checkpoint.ts), [todo.ts](../examples/extensions/todo.ts), [tools.ts](../examples/extensions/tools.ts)
+
 #### session_before_compact / session_compact
 
 Fired on compaction. See [compaction.md](compaction.md) for details.
@@ -376,6 +382,8 @@ pi.on("session_compact", async (event, ctx) => {
 });
 ```
 
+**Examples:** [custom-compaction.ts](../examples/extensions/custom-compaction.ts)
+
 #### session_before_tree / session_tree
 
 Fired on `/tree` navigation.
@@ -393,6 +401,8 @@ pi.on("session_tree", async (event, ctx) => {
 });
 ```
 
+**Examples:** [todo.ts](../examples/extensions/todo.ts), [tools.ts](../examples/extensions/tools.ts)
+
 #### session_shutdown
 
 Fired on exit (Ctrl+C, Ctrl+D, SIGTERM).
@@ -403,16 +413,19 @@ pi.on("session_shutdown", async (_event, ctx) => {
 });
 ```
 
+**Examples:** [auto-commit-on-exit.ts](../examples/extensions/auto-commit-on-exit.ts)
+
 ### Agent Events
 
 #### before_agent_start
 
-Fired after user submits prompt, before agent loop. Can inject a message and/or append to the system prompt.
+Fired after user submits prompt, before agent loop. Can inject a message and/or modify the system prompt.
 
 ```typescript
 pi.on("before_agent_start", async (event, ctx) => {
 	// event.prompt - user's prompt text
 	// event.images - attached images (if any)
+	// event.systemPrompt - current system prompt
 
 	return {
 		// Inject a persistent message (stored in session, sent to LLM)
@@ -421,12 +434,14 @@ pi.on("before_agent_start", async (event, ctx) => {
 			content: "Additional context for the LLM",
 			display: true,
 		},
-		// Append to system prompt for this turn only
-		systemPromptAppend: "Extra instructions for this turn...",
+		// Replace the system prompt for this turn (chained across extensions)
+		systemPrompt: event.systemPrompt + "\n\nExtra instructions for this turn...",
 	};
 });
 ```
 
+**Examples:** [claude-rules.ts](../examples/extensions/claude-rules.ts), [pirate.ts](../examples/extensions/pirate.ts), [plan-mode.ts](../examples/extensions/plan-mode.ts), [preset.ts](../examples/extensions/preset.ts), [ssh.ts](../examples/extensions/ssh.ts)
+
 #### agent_start / agent_end
 
 Fired once per user prompt.
@@ -439,6 +454,8 @@ pi.on("agent_end", async (event, ctx) => {
 });
 ```
 
+**Examples:** [chalk-logger.ts](../examples/extensions/chalk-logger.ts), [git-checkpoint.ts](../examples/extensions/git-checkpoint.ts), [plan-mode.ts](../examples/extensions/plan-mode.ts)
+
 #### turn_start / turn_end
 
 Fired for each turn (one LLM response + tool calls).
@@ -453,6 +470,8 @@ pi.on("turn_end", async (event, ctx) => {
 });
 ```
 
+**Examples:** [git-checkpoint.ts](../examples/extensions/git-checkpoint.ts), [plan-mode.ts](../examples/extensions/plan-mode.ts), [status-line.ts](../examples/extensions/status-line.ts)
+
 #### context
 
 Fired before each LLM call. Modify messages non-destructively.
@@ -465,6 +484,8 @@ pi.on("context", async (event, ctx) => {
 });
 ```
 
+**Examples:** [plan-mode.ts](../examples/extensions/plan-mode.ts)
+
 ### Tool Events
 
 #### tool_call
@@ -483,6 +504,8 @@ pi.on("tool_call", async (event, ctx) => {
 });
 ```
 
+**Examples:** [chalk-logger.ts](../examples/extensions/chalk-logger.ts), [permission-gate.ts](../examples/extensions/permission-gate.ts), [plan-mode.ts](../examples/extensions/plan-mode.ts), [protected-paths.ts](../examples/extensions/protected-paths.ts)
+
 #### tool_result
 
 Fired after tool executes. **Can modify result.**
@@ -503,6 +526,8 @@ pi.on("tool_result", async (event, ctx) => {
 });
 ```
 
+**Examples:** [git-checkpoint.ts](../examples/extensions/git-checkpoint.ts), [plan-mode.ts](../examples/extensions/plan-mode.ts)
+
 ## ExtensionContext
 
 Every handler receives `ctx: ExtensionContext`:

package/docs/sdk.md

@@ -532,7 +532,7 @@ Hook API methods:
 import { createAgentSession, discoverSkills, type Skill } from "@oh-my-pi/pi-coding-agent";
 
 // Discover and filter
-const allSkills = discoverSkills();
+const { skills: allSkills, warnings } = discoverSkills();
 const filtered = allSkills.filter((s) => s.name.includes("search"));
 
 // Custom skill
@@ -554,7 +554,7 @@ const { session } = await createAgentSession({
 });
 
 // Discovery with settings filter
-const skills = discoverSkills(process.cwd(), undefined, {
+const { skills } = discoverSkills(process.cwd(), undefined, {
 	ignoredSkills: ["browser-*"], // glob patterns to exclude
 	includeSkills: ["search-*"], // glob patterns to include (empty = all)
 });
@@ -753,7 +753,7 @@ const model = modelRegistry.find("provider", "id"); // Find specific model
 const builtIn = getModel("anthropic", "claude-opus-4-5"); // Built-in only
 
 // Skills
-const skills = discoverSkills(cwd, agentDir, skillsSettings);
+const { skills, warnings } = discoverSkills(cwd, agentDir, skillsSettings);
 
 // Hooks (async - loads TypeScript)
 const hooks = await discoverHooks(cwd, agentDir);

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@oh-my-pi/pi-coding-agent",
-	"version": "3.37.0",
+	"version": "4.0.0",
 	"description": "Coding agent CLI with read, bash, edit, write tools and session management",
 	"type": "module",
 	"ompConfig": {
@@ -39,10 +39,10 @@
 		"prepublishOnly": "bun run generate-template && bun run clean && bun run build"
 	},
 	"dependencies": {
-		"@oh-my-pi/pi-ai": "3.37.0",
-		"@oh-my-pi/pi-agent-core": "3.37.0",
-		"@oh-my-pi/pi-git-tool": "3.37.0",
-		"@oh-my-pi/pi-tui": "3.37.0",
+		"@oh-my-pi/pi-ai": "4.0.0",
+		"@oh-my-pi/pi-agent-core": "4.0.0",
+		"@oh-my-pi/pi-git-tool": "4.0.0",
+		"@oh-my-pi/pi-tui": "4.0.0",
 		"@openai/agents": "^0.3.7",
 		"@sinclair/typebox": "^0.34.46",
 		"ajv": "^8.17.1",

package/src/cli/args.ts

@@ -30,8 +30,10 @@ export interface Args {
 	sessionDir?: string;
 	models?: string[];
 	tools?: string[];
+	noTools?: boolean;
 	hooks?: string[];
 	extensions?: string[];
+	noExtensions?: boolean;
 	print?: boolean;
 	export?: string;
 	noSkills?: boolean;
@@ -96,6 +98,8 @@ export function parseArgs(args: string[], extensionFlags?: Map<string, { type: "
 			result.sessionDir = args[++i];
 		} else if (arg === "--models" && i + 1 < args.length) {
 			result.models = args[++i].split(",").map((s) => s.trim());
+		} else if (arg === "--no-tools") {
+			result.noTools = true;
 		} else if (arg === "--tools" && i + 1 < args.length) {
 			const toolNames = args[++i].split(",").map((s) => s.trim());
 			const validTools: string[] = [];
@@ -132,6 +136,8 @@ export function parseArgs(args: string[], extensionFlags?: Map<string, { type: "
 		} else if ((arg === "--extension" || arg === "-e") && i + 1 < args.length) {
 			result.extensions = result.extensions ?? [];
 			result.extensions.push(args[++i]);
+		} else if (arg === "--no-extensions") {
+			result.noExtensions = true;
 		} else if (arg === "--no-skills") {
 			result.noSkills = true;
 		} else if (arg === "--skills" && i + 1 < args.length) {
@@ -189,11 +195,13 @@ ${chalk.bold("Options:")}
   --no-session                   Don't save session (ephemeral)
   --models <patterns>            Comma-separated model patterns for Ctrl+P cycling
                                  Supports globs (anthropic/*, *sonnet*) and fuzzy matching
+  --no-tools                     Disable all built-in tools
   --tools <tools>                Comma-separated list of tools to enable (default: read,bash,edit,write)
                                  Available: read, bash, edit, write, grep, find, ls
   --thinking <level>             Set thinking level: off, minimal, low, medium, high, xhigh
   --hook <path>                  Load a hook/extension file (can be used multiple times)
   --extension, -e <path>         Load an extension file (can be used multiple times)
+  --no-extensions                Disable extension discovery (explicit -e paths still work)
   --no-skills                    Disable skills discovery and loading
   --skills <patterns>            Comma-separated glob patterns to filter skills (e.g., git-*,docker)
   --export <file>                Export session file to HTML and exit

package/src/config.ts

@@ -45,21 +45,6 @@ export function getPackageDir(): string {
 	return process.cwd();
 }
 
-/** Get path to README.md (optional, may not exist in binary) */
-export function getReadmePath(): string {
-	return resolve(join(getPackageDir(), "README.md"));
-}
-
-/** Get path to docs directory (optional, may not exist in binary) */
-export function getDocsPath(): string {
-	return resolve(join(getPackageDir(), "docs"));
-}
-
-/** Get path to examples directory (optional, may not exist in binary) */
-export function getExamplesPath(): string {
-	return resolve(join(getPackageDir(), "examples"));
-}
-
 /** Get path to CHANGELOG.md (optional, may not exist in binary) */
 export function getChangelogPath(): string {
 	return resolve(join(getPackageDir(), "CHANGELOG.md"));
@@ -99,6 +84,11 @@ export function getToolsDir(): string {
 	return join(getAgentDir(), "tools");
 }
 
+/** Get path to managed binaries directory (fd, rg) */
+export function getBinDir(): string {
+	return join(getAgentDir(), "bin");
+}
+
 /** Get path to slash commands directory */
 export function getCommandsDir(): string {
 	return join(getAgentDir(), "commands");