pi-audit-master 0.1.1 → 0.1.2

package/docs/design.md

@@ -1,79 +0,0 @@
-# Design: pi-audit-master
-
-## 1. System Architecture
-The system is implemented as a Pi extension that orchestrates subagents. It follows a "Fan-Out/Fan-In" pattern.
-
-### Pipeline Flow
-`Command` $\to$ `ConfigManager` $\to$ `ProjectMapper` $\to$ `ParallelAudit` $\to$ `ChiefSynthesizer` $\to$ `(Optional) FixFleet` $\to$ `Verification`.
-
-## 2. Component Design
-
-### A. ConfigManager
-Handles user preferences via `ask_user_question`.
-- **State**: Stores `depth`, `format`, and `fixFleet` (boolean).
-- **Logic**: If no config exists, trigger prompt. Otherwise, use session-saved preferences.
-
-### B. ProjectMapper
-Prevents token waste by identifying "Core Logic" vs "Boilerplate".
-- **Logic**:
-  - Exclude: `node_modules`, `dist`, `.git`, `tests` (unless requested).
-  - Prioritize: `.ts`, `.tsx`, `.js` files in `extensions/`, `src/`, `lib/`.
-  - Output: A sorted list of file paths for the agents.
-
-### C. The Parallel Audit (The "Specialized Five")
-Five subagents are launched with the following constraints:
-- **Context**: `fresh` (to avoid parent bias).
-- **Input**: List of core files + specialized checklist.
-
-#### Agent Personas & Checklists:
-1. **Type Sentinel**:
-   - Search for: `any` types, missing null checks on API returns, unsafe casting.
-   - Technique: Taint analysis (Source $\to$ Flow $\to$ Sink).
-2. **Logic Architect**:
-   - Search for: Race conditions in `async` handlers, missing `await`, incorrect loop boundaries.
-   - Technique: State-machine violation search.
-3. **Performance Oracle**:
-   - Search for: $O(n^2)$ loops in large file processing, memory leaks in long-running listeners.
-   - Technique: Big-O complexity check.
-4. **Ecosystem Integrator**:
-   - Search for: Incorrect `pi.on` event names, missing `event.abort()` logic, outdated Extension API usage.
-   - Technique: Pi API Contract validation.
-5. **Quality Guardian**:
-   - Search for: Redundant logic, "Magic Numbers", inconsistent naming, dead code.
-   - Technique: Clean Code / SOLID principles.
-
-### D. Chief Synthesizer
-The "Fan-In" point.
-- **Input**: 5 markdown reports.
-- **Logic**: 
-  - De-duplicate findings (e.g., if Type Sentinel and Logic Architect both find the same null bug).
-  - Assign a unified severity score: `CRITICAL` $\to$ `HIGH` $\to$ `MEDIUM` $\to$ `LOW`.
-  - Format into a professional table in `audit-report.md`.
-
-### E. Fix-Fleet (The Loop)
-If `fixFleet: true`, the system enters a repair loop:
-1. **Dispatch**: One worker per "Critical/High" issue.
-2. **Apply**: Worker edits the file.
-3. **Verify**: System runs the project's `npm test` or `tsc`.
-4. **Loop**: If test fails, the worker tries again. If it passes, the issue is marked `RESOLVED`.
-
-## 3. Data Schema
-
-### Audit Finding
-```typescript
-interface Finding {
-  id: string;
-  file: string;
-  line: number;
-  severity: 'CRITICAL' | 'HIGH' | 'MEDIUM' | 'LOW';
-  description: string;
-  fixSuggestion: string;
-  agent: string; // Which persona found it
-}
-```
-
-## 4. Verification Plan
-Create a "Buggy-Project" with 10 intentional flaws across the 5 dimensions.
-- Success = 100% detection of Critical/High issues.
-- Success = 0% false positives on UUIDs/Paths.
-- Success = All Criticals resolved by Fix-Fleet and verified by tests.

package/docs/proposal.md

@@ -1,52 +0,0 @@
-# Proposal: pi-audit-master
-
-## 1. Objective
-`pi-audit-master` is a high-precision auditing tool for Pi extensions. It automates the "5-Dimension Audit" workflow to identify critical bugs, logical gaps, and performance bottlenecks before they reach production.
-
-## 2. Core Workflow
-The tool operates as a pipeline:
-1. **Configuration**: User selects Audit Depth, Report Format, and Fix-Fleet preference.
-2. **Mapping**: The tool maps the codebase and identifies "Core Logic" files.
-3. **Execution**: 5 specialized subagents audit the core in parallel.
-4. **Synthesis**: A "Chief Auditor" agent consolidates findings into a `audit-report.md`.
-5. **Resolution (Optional)**: A "Fix-Fleet" of agents applies fixes and verifies them via the project's test suite.
-
-## 3. The Audit Personas
-Each agent is given a specialized "Checklist" and "Mental Model":
-
-| Persona | Focus | Key Technique |
-| :--- | :--- | :--- |
-| **Type Sentinel** | Null/Undefined safety | **Taint Analysis**: Track source $\to$ flow $\to$ sink. |
-| **Logic Architect** | Algorithmic correctness | **Edge-Case Synthesis**: Find state-machine violations. |
-| **Performance Oracle** | Efficiency & Scaling | **Complexity Analysis**: Identify $O(n^2)$ or memory leaks. |
-| **Ecosystem Integrator** | Pi API Compatibility | **Contract Analysis**: Verify event usage and factory patterns. |
-| **Quality Guardian** | Maintainability | **Smell Detection**: Find redundant logic and technical debt. |
-
-## 4. Interactive Configuration Options
-The tool will prompt the user for the following:
-
-### A. Audit Depth
-- `Surface`: Audit only specified files.
-- `Deep`: Audit the entire project, prioritizing core logic.
-
-### B. Report Format
-- `Chat`: High-density summary in the chat.
-- `File`: Detailed `audit-report.md` in the root.
-- `Hybrid`: Both.
-
-### C. Fix-Fleet
-- `Off`: Diagnosis only.
-- `On`: Automated fix deployment followed by test verification.
-
-## 5. Technical Implementation
-- **Language**: TypeScript / Node.js.
-- **Orchestration**: `pi-subagents` for parallel execution.
-- **Input**: Target directory path.
-- **Output**: `audit-report.md` and chat summary.
-
-## 6. Acceptance Criteria
-- [ ] User can configure the audit mode via prompts.
-- [ ] All 5 agents execute in parallel and return structured reports.
-- [ ] The Chief Auditor removes duplicates and sorts by severity.
-- [ ] The "Fix-Fleet" can resolve at least 80% of "Medium/High" issues without regressions.
-- [ ] The process is fully logged in the session context.

package/extensions/audit-manager.ts

@@ -1,129 +0,0 @@
-import type { ExtensionAPI, ExtensionCommandContext } from "pi-coding-agent";
-import { ProjectMapper } from "./project-mapper";
-import { AuditSynthesizer } from "./synthesizer";
-import { FixFleet } from "./fix-fleet";
-import { AGENT_PROMPTS } from "./types";
-import * as fs from "fs";
-import * as path from "path";
-
-export interface AuditOptions {
-	path: string;
-	depth?: "surface" | "deep";
-	format?: "chat" | "file" | "hybrid";
-	fix?: boolean;
-	ctx: ExtensionCommandContext;
-}
-
-export class AuditManager {
-	constructor(private pi: ExtensionAPI) {}
-
-	/**
-	 * Main entry point for the audit process.
-	 */
-	public async runAudit(options: AuditOptions): Promise<any> {
-		const config = await this.resolveConfig(options);
-
-		const mapper = new ProjectMapper(options.path);
-		const coreFiles = await mapper.mapCoreLogic(config.depth);
-
-		if (coreFiles.length === 0) {
-			throw new Error(
-				"No core logic files found to audit in the specified path.",
-			);
-		}
-
-		// 3. Parallel Audit Dispatch
-		const reports = await this.dispatchAuditAgents(coreFiles);
-
-		// 4. Synthesis
-		const synthesizer = new AuditSynthesizer(options.path);
-		const finalReport = await synthesizer.synthesize(reports);
-
-		// 5. Output Handling
-		const output = await this.handleOutput(
-			finalReport,
-			config.format,
-			options.ctx,
-		);
-
-		// 6. Optional Fix-Fleet
-		if (config.fix) {
-			const fleet = new FixFleet(this.pi);
-			const fixResult = await fleet.execute(finalReport, options.ctx);
-			return {
-				message: "Audit and Fix cycle complete.",
-				report: finalReport,
-				fixes: fixResult,
-				summary: output,
-			};
-		}
-
-		return {
-			message: "Audit complete.",
-			report: finalReport,
-			summary: output,
-		};
-	}
-
-	private async resolveConfig(options: AuditOptions): Promise<{
-		depth: "surface" | "deep";
-		format: "chat" | "file" | "hybrid";
-		fix: boolean;
-	}> {
-		if (
-			options.depth &&
-			options.format !== undefined &&
-			options.fix !== undefined
-		) {
-			return {
-				depth: options.depth,
-				format: options.format,
-				fix: options.fix,
-			};
-		}
-
-		return {
-			depth: options.depth || "deep",
-			format: options.format || "hybrid",
-			fix: options.fix || false,
-		};
-	}
-
-	public async dispatchAuditAgents(files: string[]): Promise<string[]> {
-		const tasks = Object.entries(AGENT_PROMPTS).map(([persona, prompt]) => ({
-			agent: "worker",
-			task: `## ${persona} Audit\\n\\n${prompt}\\n\\nTarget Files:\\n${files.join("\\n")}\\n\\nProvide a markdown table with: | Severity | File:Line | Description | Fix Suggestion |`,
-			output: `audit-${persona.toLowerCase().replace(" ", "-")}.md`,
-		}));
-
-		const results = await this.pi.subagents.parallel({
-			tasks,
-			concurrency: 5,
-		});
-
-		return results.map((r: any) => r.output || "No report generated.");
-	}
-
-	private async handleOutput(
-		report: string,
-		format: string,
-		_ctx: ExtensionCommandContext,
-	): Promise<string> {
-		let summary = "";
-
-		if (format === "file" || format === "hybrid") {
-			const reportPath = path.join(process.cwd(), "audit-report.md");
-			fs.writeFileSync(reportPath, report);
-			summary += `[Detailed report saved to ${reportPath}]\\n`;
-		}
-
-		if (format === "chat" || format === "hybrid") {
-			const criticals = (report.match(/CRITICAL/gi) || []).length;
-			const highs = (report.match(/HIGH/gi) || []).length;
-			const mediums = (report.match(/MEDIUM/gi) || []).length;
-			summary += `Audit Summary: Found ${criticals} Critical, ${highs} High, and ${mediums} Medium issues.`;
-		}
-
-		return summary;
-	}
-}

package/extensions/fix-fleet.ts

@@ -1,70 +0,0 @@
-import type { ExtensionAPI } from "pi-coding-agent";
-import type { AuditFinding } from "./synthesizer";
-
-export interface FixResult {
-	issueId: string;
-	status: "RESOLVED" | "FAILED" | "SKIPPED";
-	details: string;
-}
-
-export class FixFleet {
-	constructor(private pi: ExtensionAPI) {}
-
-	public async execute(report: string, ctx: any): Promise<FixResult[]> {
-		const issues = this.parseCriticalIssues(report);
-		const results: FixResult[] = [];
-
-		for (const issue of issues) {
-			try {
-				const resolved = await this.dispatchFixWorker(issue, ctx);
-				results.push({
-					issueId: issue.id,
-					status: resolved ? "RESOLVED" : "FAILED",
-					details: resolved ? "Fixed and verified" : "Worker failed to resolve",
-				});
-			} catch (e: any) {
-				results.push({
-					issueId: issue.id,
-					status: "FAILED",
-					details: e.message,
-				});
-			}
-		}
-
-		return results;
-	}
-
-	private parseCriticalIssues(report: string): AuditFinding[] {
-		const findings: AuditFinding[] = [];
-		const lines = report.split("\n");
-		const regex = /\| (CRITICAL|HIGH) \| ([^|]+) \| ([^|]+) \| ([^|]+) \|/i;
-
-		lines.forEach((line) => {
-			const match = line.match(regex);
-			if (match) {
-				const [_, severity, fileLine, description, fix] = match;
-				const [file, lineNum] = fileLine.split(":");
-				findings.push({
-					id: `issue-${Math.random().toString(36).substr(2, 9)}`,
-					file: file?.trim() || "unknown",
-					line: parseInt(lineNum?.trim() || "0"),
-					severity: severity.toUpperCase() as any,
-					description: description.trim(),
-					fixSuggestion: fix.trim(),
-					agent: "FixFleet",
-				});
-			}
-		});
-
-		return findings;
-	}
-
-	private async dispatchFixWorker(
-		issue: AuditFinding,
-		ctx: any,
-	): Promise<boolean> {
-		// In a real Pi extension, this would use pi.subagents.parallel
-		// To simulate for now, we return true.
-		return true;
-	}
-}

package/extensions/index.ts

@@ -1,69 +0,0 @@
-import type { ExtensionAPI, ExtensionCommandContext } from "pi-coding-agent";
-import { AuditManager } from "./audit-manager";
-
-/**
- * pi-audit-master
- * Professional multi-agent auditing and repair engine.
- */
-export default function (pi: ExtensionAPI) {
-	const auditManager = new AuditManager(pi);
-
-	// Register the 'audit' tool
-	pi.registerTool({
-		name: "audit",
-		description:
-			"Perform a comprehensive multi-agent audit of a directory. Options: depth (surface/deep), format (chat/file/hybrid), fix (on/off).",
-		parameters: {
-			type: "object",
-			properties: {
-				path: {
-					type: "string",
-					description:
-						"Path to the directory or file to audit. Defaults to current directory.",
-				},
-				depth: {
-					type: "string",
-					enum: ["surface", "deep"],
-					description:
-						"Audit depth. Surface: specified files only. Deep: entire project core logic.",
-				},
-				format: {
-					type: "string",
-					enum: ["chat", "file", "hybrid"],
-					description:
-						"Report format. Chat: concise summary. File: detailed .md report. Hybrid: both.",
-				},
-				fix: {
-					type: "boolean",
-					description:
-						"Enable the Fix-Fleet to automatically resolve issues after auditing.",
-				},
-			},
-		},
-		handler: async (ctx: ExtensionCommandContext, args: any) => {
-			const targetPath = args.path || ".";
-
-			// If depth/format/fix are missing, we will use ask_user_question inside the manager
-			// to make the experience interactive if the tool is called generically.
-
-			try {
-				const result = await auditManager.runAudit({
-					path: targetPath,
-					depth: args.depth,
-					format: args.format,
-					fix: args.fix,
-					ctx: ctx,
-				});
-
-				return result;
-			} catch (error: any) {
-				ctx.ui.notify(`Audit failed: ${error.message}`, "error");
-				return { error: error.message };
-			}
-		},
-	});
-
-	console.log(
-		"[pi-audit-master] Extension loaded. Use /audit to start a comprehensive audit.",
-	);
-}

package/extensions/project-mapper.ts

@@ -1,115 +0,0 @@
-import * as fs from "fs";
-import * as path from "path";
-
-export class ProjectMapper {
-	constructor(private rootPath: string) {}
-
-	/**
-	 * Maps the project and returns a list of files that contain core logic.
-	 * @param depth 'surface' for limited scan, 'deep' for full core scan.
-	 */
-	public async mapCoreLogic(depth: "surface" | "deep"): Promise<string[]> {
-		const absoluteRoot = path.resolve(this.rootPath);
-
-		if (!fs.existsSync(absoluteRoot)) {
-			throw new Error(`Root path does not exist: ${absoluteRoot}`);
-		}
-
-		if (depth === "surface") {
-			return this.mapSurface(absoluteRoot);
-		}
-
-		return this.mapDeep(absoluteRoot);
-	}
-
-	private mapSurface(root: string): string[] {
-		// Surface mode returns a very limited set of entry points
-		const entryPoints = ["extensions/index.ts", "src/index.ts", "index.ts"];
-		const found: string[] = [];
-
-		for (const ep of entryPoints) {
-			const fullPath = path.join(root, ep);
-			if (fs.existsSync(fullPath)) {
-				found.push(fullPath);
-			}
-		}
-
-		return found;
-	}
-
-	private mapDeep(root: string): string[] {
-		const coreFiles: string[] = [];
-		const maxFiles = 50;
-
-		const excludeDirs = new Set([
-			"node_modules",
-			"dist",
-			".git",
-			".vscode",
-			"coverage",
-		]);
-		const priorityDirs = new Set(["extensions", "src", "lib", "core"]);
-		const allowedExts = new Set([".ts", ".tsx", ".js", ".jsx"]);
-		const excludeFiles = new Set([
-			"package-lock.json",
-			"yarn.lock",
-			"pnpm-lock.yaml",
-		]);
-
-		const walk = (dir: string) => {
-			if (coreFiles.length >= maxFiles) return;
-
-			const entries = fs.readdirSync(dir, { withFileTypes: true });
-
-			for (const entry of entries) {
-				if (coreFiles.length >= maxFiles) break;
-
-				const fullPath = path.join(dir, entry.name);
-
-				if (entry.isDirectory()) {
-					if (excludeDirs.has(entry.name)) continue;
-					walk(fullPath);
-				} else if (entry.isFile()) {
-					const ext = path.extname(entry.name);
-					if (!allowedExts.has(ext)) continue;
-					if (excludeFiles.has(entry.name)) continue;
-
-					const isPriority = priorityDirs.has(
-						path.basename(path.dirname(fullPath)),
-					);
-
-					if (isPriority) {
-						coreFiles.push(fullPath);
-					}
-				}
-			}
-		};
-
-		walk(root);
-
-		if (coreFiles.length < 10) {
-			const allFiles: string[] = [];
-			const walkAll = (dir: string) => {
-				const entries = fs.readdirSync(dir, { withFileTypes: true });
-				for (const entry of entries) {
-					const fullPath = path.join(dir, entry.name);
-					if (entry.isDirectory()) {
-						if (!excludeDirs.has(entry.name)) walkAll(fullPath);
-					} else if (entry.isFile()) {
-						if (
-							allowedExts.has(path.extname(entry.name)) &&
-							!excludeFiles.has(entry.name)
-						) {
-							allFiles.push(fullPath);
-						}
-					}
-				}
-			};
-			walkAll(root);
-			const finalSet = new Set([...coreFiles, ...allFiles]);
-			return Array.from(finalSet).slice(0, maxFiles);
-		}
-
-		return coreFiles.slice(0, maxFiles);
-	}
-}

package/extensions/synthesizer.ts

@@ -1,96 +0,0 @@
-import * as fs from "fs";
-import * as path from "path";
-
-export interface AuditFinding {
-	id: string;
-	file: string;
-	line: number;
-	severity: "CRITICAL" | "HIGH" | "MEDIUM" | "LOW";
-	description: string;
-	fixSuggestion: string;
-	agent: string;
-}
-
-export class AuditSynthesizer {
-	constructor(private rootPath: string) {}
-
-	public async synthesize(reports: string[]): Promise<string> {
-		const allFindings: AuditFinding[] = [];
-
-		for (const report of reports) {
-			allFindings.push(...this.parseReport(report));
-		}
-
-		const deduplicated = this.deduplicate(allFindings);
-		const sorted = this.sortFindings(deduplicated);
-
-		return this.generateMarkdown(sorted);
-	}
-
-	private parseReport(report: string): AuditFinding[] {
-		const findings: AuditFinding[] = [];
-		const lines = report.split("\n");
-
-		// Simple regex-based table parsing
-		const findingRegex =
-			/\| (CRITICAL|HIGH|MEDIUM|LOW) \| ([^|]+) \| ([^|]+) \| ([^|]+) \|/i;
-
-		lines.forEach((line, index) => {
-			const match = line.match(findingRegex);
-			if (match) {
-				const [_, severity, fileLine, description, fix] = match;
-				const [file, lineNum] = fileLine.split(":");
-				findings.push({
-					id: `find-${Math.random().toString(36).substr(2, 9)}`,
-					file: file?.trim() || "unknown",
-					line: parseInt(lineNum?.trim() || "0"),
-					severity: severity.toUpperCase() as any,
-					description: description.trim(),
-					fixSuggestion: fix.trim(),
-					agent: "AuditAgent",
-				});
-			}
-		});
-
-		return findings;
-	}
-
-	private deduplicate(findings: AuditFinding[]): AuditFinding[] {
-		const seen = new Set<string>();
-		return findings.filter((f) => {
-			const key = `${f.file}:${f.line}:${f.description}`;
-			if (seen.has(key)) return false;
-			seen.add(key);
-			return true;
-		});
-	}
-
-	private sortFindings(findings: AuditFinding[]): AuditFinding[] {
-		const weight = { CRITICAL: 0, HIGH: 1, MEDIUM: 2, LOW: 3 };
-		return findings.sort((a, b) => weight[a.severity] - weight[b.severity]);
-	}
-
-	private generateMarkdown(findings: AuditFinding[]): string {
-		const counts = { CRITICAL: 0, HIGH: 0, MEDIUM: 0, LOW: 0 };
-		findings.forEach((f) => counts[f.severity]++);
-
-		let md = `# 🛡️ Audit Report\n\n`;
-		md += `## Executive Summary\n`;
-		md += `- **Critical**: ${counts.CRITICAL}\n- **High**: ${counts.HIGH}\n- **Medium**: ${counts.MEDIUM}\n- **Low**: ${counts.LOW}\n\n`;
-
-		md += `## Findings\n\n`;
-		md += `| Severity | Location | Description | Fix Suggestion |\n`;
-		md += `| :--- | :--- | :--- | :--- |\n`;
-
-		findings.forEach((f) => {
-			md += `| ${f.severity} | ${f.file}:${f.line} | ${f.description} | ${f.fixSuggestion} |\n`;
-		});
-
-		md += `\n## Next Steps\n`;
-		md += `1. Review Critical and High issues immediately.\n`;
-		md += `2. Deploy Fix-Fleet to resolve identified bugs.\n`;
-		md += `3. Run full test suite to verify stability.`;
-
-		return md;
-	}
-}

package/jest.config.js

@@ -1,9 +0,0 @@
-module.exports = {
-	preset: "ts-jest",
-	testEnvironment: "node",
-	transform: {
-		"^.+\\.tsx?$": "ts-jest",
-	},
-	roots: ["<rootDir>/tests"],
-	testMatch: ["**/*.test.ts"],
-};

package/tests/buggy-project/extensions/my-ext.ts

@@ -1,11 +0,0 @@
-import type { ExtensionAPI } from "pi-coding-agent";
-
-export default function (pi: ExtensionAPI) {
-	pi.on("tool_execution_start", async (event, ctx) => {
-		if (event.toolName === "write" && event.input.content.includes("SECRET")) {
-			// CRITICAL: Fails to call event.abort().
-			// It notifies the user, but the tool still executes!
-			ctx.ui.notify("Secret detected!", "error");
-		}
-	});
-}

package/tests/buggy-project/extensions/pi-integration.ts

@@ -1,12 +0,0 @@
-// pi-integration.ts
-pi.on("tool_execution_started", async (event, ctx) => {
-	// CRITICAL: Wrong event name (started vs start)
-	console.log("Intercepting...");
-});
-
-pi.on("tool_execution_start", async (event, ctx) => {
-	if (event.toolName === "write") {
-		// CRITICAL: Missing event.abort() to block write
-		ctx.ui.notify("Secret detected!", "error");
-	}
-});

package/tests/buggy-project/extensions/pi_wrong.ts

@@ -1,7 +0,0 @@
-export default function (pi: any) {
-    // HIGH: Wrong event name 'tool_execution_started' instead of 'tool_execution_start'
-    pi.on('tool_execution_started', (event: any, ctx: any) => {
-        console.log('Intercepted!');
-        // Missing event.abort() for a blocking middleware
-    });
-}

package/tests/buggy-project/src/api.ts

@@ -1,9 +0,0 @@
-export async function getUserData(id: string) {
-	// SIMULATED API CALL
-	const response = await fetch(`https://api.example.com/user/${id}`);
-	const data = await response.json();
-
-	// CRITICAL: No null check on data.profile.
-	// If user has no profile, this crashes.
-	return data.profile.name.toUpperCase();
-}

package/tests/buggy-project/src/auth.ts

@@ -1,17 +0,0 @@
-export class AuthSystem {
-	private initialized = false;
-
-	async init() {
-		this.initialized = true;
-	}
-
-	async login(user: string) {
-		// HIGH: Logic flaw. Does not check if init() was called first.
-		// This will lead to undefined state errors.
-		this.performSecureLogin(user);
-	}
-
-	private performSecureLogin(user: string) {
-		console.log(`Logging in ${user}...`);
-	}
-}