SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsproject-sanitizer

project-sanitizer

Office & productivity

View repository

Updated 10-15 23:58

About this skill

Project Sanitizer & Analysis Assistant

Overview

This skill provides a workflow for reviewing projects that contain reverse-engineering artifacts. It assists in the initial analysis of binaries and automates the subsequent redaction pass before internal review or public sharing.

Workflow

Phase 1: Analysis & Recovery (Optional)

Use this phase to extract core information needed for "narrative reconstruction".

# Analyze a binary to extract strings/symbols
./local-skills/project-sanitizer/scripts/analyze_target.sh ./binaries/my-app

Methodology guidance lives in references/:

  • references/analysis-methodology.md — eight transferable techniques (function fingerprinting, XREF chokepoint heuristic, Swift/ObjC metadata exploitation, dynamic hooking choices, backtrace-at-symptom, string xrefs incl. Swift SSO inline imm, differential analysis, observation-system awareness) plus the five-layer verification protocol and the subtleties that bite first-timers (universal binary offsets, ASLR slide arithmetic, /etc/hosts bypass by userspace DNS, codesign --deep sealed resources, Hardened Runtime vs App Sandbox).
  • references/analysis-checklist.md — condensed cheat sheet of the same material; keep open during an audit.

Read the methodology before reaching for a disassembler. Most "this binary behaves weirdly" tickets resolve at the configuration layer; binary inspection is for when cheap layers cannot explain the behaviour.

Phase 2: Sanitization (Essential)

Run this phase before sharing with Claude Code or uploading to GitHub.

# Scrub RE traces and rebrand
./local-skills/project-sanitizer/scripts/sanitize_project.sh .

# Preview changes without modifying files
./local-skills/project-sanitizer/scripts/sanitize_project.sh --dry-run .

Phase 3: Provenance Review & Upload

  1. Update README.md to accurately describe the project provenance and remaining third-party material.
  2. Initialize a fresh Git repository.
  3. Push to GitHub using gh.

Resources

scripts/sanitize_project.sh

  • Deletes analysis/, binaries/, pseudocode/, etc.
  • Recursively deletes lines containing cracking fingerprints (e.g., CRACKED, fake_, 破解).
  • Globally replaces RE-related terms with neutral identifiers (e.g., 反汇编 -> 源码分析).

scripts/analyze_target.sh

  • Extracts strings, exported symbols, and library dependencies from a target binary.
  • Outputs results to an analysis/ folder (which is later scrubbed by the sanitizer).

references/analysis-methodology.md

  • Long-form methodology for binary analysis (security audit, legacy archaeology, third-party SDK due diligence, crash investigation).
  • Covers when to start binary work vs. exhaust configuration-layer investigation, the eight technique catalogue, five-layer verification protocol, and common subtleties.

references/analysis-checklist.md

  • Operational cheat sheet for an analyst doing an audit: initial recon commands, function-finding decision tree, patch verification matrix, Mach-O section reference, universal-binary offset arithmetic.