co-mono/packages/coding-agent/examples/hooks/plan-mode.ts
Helmut Januschka 537d672f17 feat(plan-mode): add todo list extraction and progress tracking
- Extract numbered steps from agent's plan response
- Track progress during execution with footer indicator (📋 2/5)
- /todos command to view current plan progress
- State persists across sessions including todo progress
- Agent prompted to format plans as numbered lists for tracking
2026-01-04 18:13:28 +01:00

470 lines
14 KiB
TypeScript

/**
* Plan Mode Hook
*
* Provides a Claude Code-style "plan mode" for safe code exploration.
* When enabled, the agent can only use read-only tools and cannot modify files.
*
* Features:
* - /plan command to toggle plan mode
* - In plan mode: only read, bash (read-only), grep, find, ls are available
* - Injects system context telling the agent about the restrictions
* - After each agent response, prompts to execute the plan or continue planning
* - Shows "plan" indicator in footer when active
* - Extracts todo list from plan and tracks progress during execution
*
* Usage:
* 1. Copy this file to ~/.pi/agent/hooks/ or your project's .pi/hooks/
* 2. Use /plan to toggle plan mode on/off
* 3. Or start in plan mode with --plan flag
*/
import type { HookAPI, HookContext } from "@mariozechner/pi-coding-agent/hooks";
// Read-only tools for plan mode
const PLAN_MODE_TOOLS = ["read", "bash", "grep", "find", "ls"];
// Full set of tools for normal mode
const NORMAL_MODE_TOOLS = ["read", "bash", "edit", "write"];
// Patterns for destructive bash commands that should be blocked in plan mode
const DESTRUCTIVE_PATTERNS = [
// File/directory modification
/\brm\b/i,
/\brmdir\b/i,
/\bmv\b/i,
/\bcp\b/i, // cp can overwrite files
/\bmkdir\b/i,
/\btouch\b/i,
/\bchmod\b/i,
/\bchown\b/i,
/\bchgrp\b/i,
/\bln\b/i, // symlinks
// File content modification
/\btee\b/i,
/\btruncate\b/i,
/\bdd\b/i,
/\bshred\b/i,
// Redirects that write to files
/[^<]>(?!>)/, // > but not >> or <>
/>>/, // append
// Package managers / installers
/\bnpm\s+(install|uninstall|update|ci|link|publish)/i,
/\byarn\s+(add|remove|install|publish)/i,
/\bpnpm\s+(add|remove|install|publish)/i,
/\bpip\s+(install|uninstall)/i,
/\bapt(-get)?\s+(install|remove|purge|update|upgrade)/i,
/\bbrew\s+(install|uninstall|upgrade)/i,
// Git write operations
/\bgit\s+(add|commit|push|pull|merge|rebase|reset|checkout\s+-b|branch\s+-[dD]|stash|cherry-pick|revert|tag|init|clone)/i,
// Other dangerous commands
/\bsudo\b/i,
/\bsu\b/i,
/\bkill\b/i,
/\bpkill\b/i,
/\bkillall\b/i,
/\breboot\b/i,
/\bshutdown\b/i,
/\bsystemctl\s+(start|stop|restart|enable|disable)/i,
/\bservice\s+\S+\s+(start|stop|restart)/i,
// Editors (interactive, could modify files)
/\b(vim?|nano|emacs|code|subl)\b/i,
];
// Read-only commands that are always safe
const SAFE_COMMANDS = [
/^\s*cat\b/,
/^\s*head\b/,
/^\s*tail\b/,
/^\s*less\b/,
/^\s*more\b/,
/^\s*grep\b/,
/^\s*find\b/,
/^\s*ls\b/,
/^\s*pwd\b/,
/^\s*echo\b/,
/^\s*printf\b/,
/^\s*wc\b/,
/^\s*sort\b/,
/^\s*uniq\b/,
/^\s*diff\b/,
/^\s*file\b/,
/^\s*stat\b/,
/^\s*du\b/,
/^\s*df\b/,
/^\s*tree\b/,
/^\s*which\b/,
/^\s*whereis\b/,
/^\s*type\b/,
/^\s*env\b/,
/^\s*printenv\b/,
/^\s*uname\b/,
/^\s*whoami\b/,
/^\s*id\b/,
/^\s*date\b/,
/^\s*cal\b/,
/^\s*uptime\b/,
/^\s*ps\b/,
/^\s*top\b/,
/^\s*htop\b/,
/^\s*free\b/,
/^\s*git\s+(status|log|diff|show|branch|remote|config\s+--get)/i,
/^\s*git\s+ls-/i,
/^\s*npm\s+(list|ls|view|info|search|outdated|audit)/i,
/^\s*yarn\s+(list|info|why|audit)/i,
/^\s*node\s+--version/i,
/^\s*python\s+--version/i,
/^\s*curl\s/i, // curl without -o is usually safe (reading)
/^\s*wget\s+-O\s*-/i, // wget to stdout only
/^\s*jq\b/,
/^\s*sed\s+-n/i, // sed with -n (no auto-print) for reading only
/^\s*awk\b/,
/^\s*rg\b/, // ripgrep
/^\s*fd\b/, // fd-find
/^\s*bat\b/, // bat (cat clone)
/^\s*exa\b/, // exa (ls clone)
];
/**
* Check if a bash command is safe (read-only) for plan mode.
*/
function isSafeCommand(command: string): boolean {
// Check if it's an explicitly safe command
if (SAFE_COMMANDS.some((pattern) => pattern.test(command))) {
// But still check for destructive patterns (e.g., cat > file)
if (!DESTRUCTIVE_PATTERNS.some((pattern) => pattern.test(command))) {
return true;
}
}
// Check for destructive patterns
if (DESTRUCTIVE_PATTERNS.some((pattern) => pattern.test(command))) {
return false;
}
// Allow commands that don't match any destructive pattern
// This is permissive - unknown commands are allowed
return true;
}
// Todo item for plan execution tracking
interface TodoItem {
text: string;
completed: boolean;
}
/**
* Extract todo items from assistant message.
* Looks for numbered lists like:
* 1. First task
* 2. Second task
* Or bullet points with step indicators:
* - Step 1: Do something
* - Step 2: Do another thing
*/
function extractTodoItems(message: string): TodoItem[] {
const items: TodoItem[] = [];
// Match numbered lists: "1. Task" or "1) Task"
const numberedPattern = /^\s*(\d+)[.)]\s+(.+)$/gm;
for (const match of message.matchAll(numberedPattern)) {
const text = match[2].trim();
// Skip if it's just a file path or code reference
if (text.length > 5 && !text.startsWith("`") && !text.startsWith("/")) {
items.push({ text, completed: false });
}
}
// If no numbered items found, try bullet points with "Step" prefix
if (items.length === 0) {
const stepPattern = /^\s*[-*]\s*(?:Step\s*\d+[:.])?\s*(.+)$/gim;
for (const match of message.matchAll(stepPattern)) {
const text = match[1].trim();
if (text.length > 10 && !text.startsWith("`")) {
items.push({ text, completed: false });
}
}
}
return items;
}
/**
* Try to match a tool call or message to a todo item.
* Returns the index of the matching item, or -1 if no match.
*/
function matchTodoItem(todos: TodoItem[], action: string): number {
const actionLower = action.toLowerCase();
for (let i = 0; i < todos.length; i++) {
if (todos[i].completed) continue;
const todoLower = todos[i].text.toLowerCase();
// Check for keyword overlap
const todoWords = todoLower.split(/\s+/).filter((w) => w.length > 3);
const matchCount = todoWords.filter((w) => actionLower.includes(w)).length;
// If more than 30% of significant words match, consider it a match
if (todoWords.length > 0 && matchCount / todoWords.length > 0.3) {
return i;
}
}
return -1;
}
export default function planModeHook(pi: HookAPI) {
// Track plan mode state
let planModeEnabled = false;
// Track execution mode (after plan confirmed)
let executionMode = false;
// Todo list extracted from plan
let todoItems: TodoItem[] = [];
// Register --plan CLI flag
pi.registerFlag("plan", {
description: "Start in plan mode (read-only exploration)",
type: "boolean",
default: false,
});
// Helper to update footer status
function updateStatus(ctx: HookContext) {
if (executionMode && todoItems.length > 0) {
const completed = todoItems.filter((t) => t.completed).length;
const total = todoItems.length;
const progress = `${completed}/${total}`;
ctx.ui.setStatus("plan-mode", ctx.ui.theme.fg("accent", `📋 ${progress}`));
} else if (planModeEnabled) {
ctx.ui.setStatus("plan-mode", ctx.ui.theme.fg("warning", "⏸ plan"));
} else {
ctx.ui.setStatus("plan-mode", undefined);
}
}
// Helper to toggle plan mode
function togglePlanMode(ctx: HookContext) {
planModeEnabled = !planModeEnabled;
executionMode = false;
todoItems = [];
if (planModeEnabled) {
pi.setTools(PLAN_MODE_TOOLS);
ctx.ui.notify(`Plan mode enabled. Tools: ${PLAN_MODE_TOOLS.join(", ")}`);
} else {
pi.setTools(NORMAL_MODE_TOOLS);
ctx.ui.notify("Plan mode disabled. Full access restored.");
}
updateStatus(ctx);
}
// Register /plan command
pi.registerCommand("plan", {
description: "Toggle plan mode (read-only exploration)",
handler: async (_args, ctx) => {
togglePlanMode(ctx);
},
});
// Register /todos command to show current todo list
pi.registerCommand("todos", {
description: "Show current plan todo list",
handler: async (_args, ctx) => {
if (todoItems.length === 0) {
ctx.ui.notify("No todos. Create a plan first with /plan", "info");
return;
}
const todoList = todoItems
.map((item, i) => {
const checkbox = item.completed ? "✓" : "○";
const style = item.completed
? ctx.ui.theme.fg("muted", `${checkbox} ${item.text}`)
: `${checkbox} ${item.text}`;
return `${i + 1}. ${style}`;
})
.join("\n");
ctx.ui.notify(`Plan Progress:\n${todoList}`, "info");
},
});
// Register Shift+P shortcut
pi.registerShortcut("shift+p", {
description: "Toggle plan mode",
handler: async (ctx) => {
togglePlanMode(ctx);
},
});
// Block destructive bash commands in plan mode
pi.on("tool_call", async (event, ctx) => {
// Track progress in execution mode
if (executionMode && todoItems.length > 0) {
const action = `${event.toolName}: ${JSON.stringify(event.input).slice(0, 200)}`;
const matchIdx = matchTodoItem(todoItems, action);
if (matchIdx >= 0) {
todoItems[matchIdx].completed = true;
updateStatus(ctx);
}
}
if (!planModeEnabled) return;
if (event.toolName !== "bash") return;
const command = event.input.command as string;
if (!isSafeCommand(command)) {
return {
block: true,
reason: `Plan mode: destructive command blocked. Use /plan to disable plan mode first.\nCommand: ${command}`,
};
}
});
// Inject plan mode context at the start of each turn via before_agent_start
pi.on("before_agent_start", async () => {
if (!planModeEnabled) return;
// Return a message to inject into context
return {
message: {
customType: "plan-mode-context",
content: `[PLAN MODE ACTIVE]
You are in plan mode - a read-only exploration mode for safe code analysis.
Restrictions:
- You can only use: read, bash, grep, find, ls
- You CANNOT use: edit, write (file modifications are disabled)
- Bash is restricted to READ-ONLY commands (cat, ls, grep, git status, etc.)
- Destructive bash commands are BLOCKED (rm, mv, cp, git commit, npm install, etc.)
- Focus on analysis, planning, and understanding the codebase
Your task is to explore, analyze, and create a detailed plan.
IMPORTANT: When you have a complete plan, format it as a numbered list:
1. First step description
2. Second step description
3. Third step description
...
This format allows tracking progress during execution.
Do NOT attempt to make changes - just describe what you would do.`,
display: false, // Don't show in TUI, just inject into context
},
};
});
// After agent finishes, offer to execute the plan
pi.on("agent_end", async (event, ctx) => {
// In execution mode, check if all todos are complete
if (executionMode && todoItems.length > 0) {
const allComplete = todoItems.every((t) => t.completed);
if (allComplete) {
ctx.ui.notify("Plan execution complete!", "info");
executionMode = false;
todoItems = [];
updateStatus(ctx);
}
return;
}
if (!planModeEnabled) return;
if (!ctx.hasUI) return;
// Try to extract todo items from the last message
const messages = event.messages;
const lastAssistant = [...messages].reverse().find((m) => m.role === "assistant");
if (lastAssistant && typeof lastAssistant.content === "string") {
const extracted = extractTodoItems(lastAssistant.content);
if (extracted.length > 0) {
todoItems = extracted;
}
}
const hasTodos = todoItems.length > 0;
const todoPreview = hasTodos
? `\n\nExtracted ${todoItems.length} steps:\n${todoItems
.slice(0, 3)
.map((t, i) => ` ${i + 1}. ${t.text.slice(0, 50)}...`)
.join("\n")}${todoItems.length > 3 ? `\n ... and ${todoItems.length - 3} more` : ""}`
: "";
const choice = await ctx.ui.select(`Plan mode - what next?${todoPreview}`, [
hasTodos ? "Execute the plan (track progress)" : "Execute the plan",
"Stay in plan mode",
"Refine the plan",
]);
if (choice?.startsWith("Execute")) {
// Switch to normal mode
planModeEnabled = false;
executionMode = hasTodos;
pi.setTools(NORMAL_MODE_TOOLS);
updateStatus(ctx);
// Send message to trigger execution immediately
const execMessage = hasTodos
? `Execute the plan you just created. There are ${todoItems.length} steps to complete. Proceed step by step, announcing each step as you work on it.`
: "Execute the plan you just created. Proceed step by step.";
pi.sendMessage(
{
customType: "plan-mode-execute",
content: execMessage,
display: true,
},
{ triggerTurn: true },
);
} else if (choice === "Refine the plan") {
const refinement = await ctx.ui.input("What should be refined?");
if (refinement) {
ctx.ui.setEditorText(refinement);
}
}
// "Stay in plan mode" - do nothing, just continue
});
// Initialize plan mode state on session start
pi.on("session_start", async (_event, ctx) => {
// Check --plan flag first
if (pi.getFlag("plan") === true) {
planModeEnabled = true;
}
// Check if there's persisted plan mode state (from previous session)
const entries = ctx.sessionManager.getEntries();
const planModeEntry = entries
.filter((e: { type: string; customType?: string }) => e.type === "custom" && e.customType === "plan-mode")
.pop() as { data?: { enabled: boolean; todos?: TodoItem[]; executing?: boolean } } | undefined;
// Restore from session (overrides flag if session has state)
if (planModeEntry?.data) {
if (planModeEntry.data.enabled !== undefined) {
planModeEnabled = planModeEntry.data.enabled;
}
if (planModeEntry.data.todos) {
todoItems = planModeEntry.data.todos;
}
if (planModeEntry.data.executing) {
executionMode = planModeEntry.data.executing;
}
}
// Apply initial state if plan mode is enabled
if (planModeEnabled) {
pi.setTools(PLAN_MODE_TOOLS);
}
updateStatus(ctx);
});
// Save state when plan mode changes (via tool_call or other events)
pi.on("turn_start", async () => {
// Persist current state including todos
pi.appendEntry("plan-mode", {
enabled: planModeEnabled,
todos: todoItems,
executing: executionMode,
});
});
}