오픈소스 · Apache 2.0 · IDE의 AI에서 동작

Due Diligence Engine

어떤 기업이 표방하는 기술은 정말 진짜일까요? DDE는 코드베이스를 읽어 기술적 주장을 검증하고 거짓과 AI-washing을 가려낸 뒤, 업계 및 경쟁사와 비교 벤치마킹하고 자사의 약점까지 점수화합니다 — 모두 IDE에서.

IDE AI Terminal — dde prompt --pdf
$ dde prompt --pdf
[STEP 0] Live web research (2026 competitors, funding, CVEs, regulations)...
[AST] Reading codebase AST, architecture & cryptographic FFI boundaries...
[RUBRIC] Scoring 5 Standard Dimensions (20% each) + Atlas 4-Axis (20/20/5/55)...
[MARKET] Building competitive landscape (7 charts × 6 markets: Global/US/EMEA/JP/SEA/LATAM)...
[MATRIX] Verifying Implementation Matrix (30 items × 10 cos., ○/△/×/? + source URLs)...
[AUDIT] Filtering Red Flags (discarding heuristic noise, keeping file:line code defects)...
✓ Score: [■■■■■■■■■■■■■■■■■■■■■■■■■■□□□□] 82/100 (Lv.8) | Grade: B
→ ~/Downloads/dde_consulting_Arc_2026-10-22.pdf (24-page consulting PDF ready)
5 Dims × 20%
Balanced Standard Score

+ Parallel Atlas 4-Axis (35% Crypto Core)

7 Charts × 6
Global Market Landscapes

+ 30-Item × 10-Co. Capability Matrix

14-Lang PDF
Consulting-Grade Output

With Data Provenance Box & Source Appendix

$0 Extra
Zero Added License Cost

Runs inside your existing IDE AI subscription

DDE란?

이미 가지고 있는 AI가, 진짜 기술과 과장을 가려냅니다

Due Diligence Engine(DDE)은 IDE의 AI 에이전트 — Claude Code, OpenAI Codex CLI, Google Gemini CLI, Antigravity, Cursor, GitHub Copilot — Claude Opus 5.5 / Sonnet 5.5, Gemini 4 Argon, GPT-5 등 — 를 세계 최고 수준의 기술 분석가로 탈바꿈시키는 오픈소스 커맨드라인 도구입니다. 포트폴리오 기업, 인수 대상, 경쟁사, 혹은 자사 팀 등 어떤 대상이든 코드베이스 전체를 읽어 표방하는 기술이 실제로 구현되었는지 검증하고, 시장과 경쟁사를 실시간으로 조사하며, 여러 가중 항목에 걸쳐 강점과 약점을 점수화하여 컨설팅 수준의 PDF 리포트를 Downloads 폴더에 작성합니다. 모든 작업은 IDE의 AI 샌드박스 안에서 로컬로 실행됩니다. API 키도, 외부 서버도, 텔레메트리도, 추가 비용도 없습니다.

From Pitch Claims to Code & Sourced Evidence

Eliminate AI-Washing: Judge Source Code and Primary Sources, Not Badges

Generic AI code reviewers stop at checkbox compliance ('Has auth? ✓ Has HTTPS? ✓'). DDE inspects actual cryptographic primitives (Signal Protocol, PQXDH + ML-KEM-1024, libsignal/BoringSSL FFI vs. self-rolled crypto), concurrency models, and live 2026 market sources.

STEP 0 · LIVE WEB PROVENANCE

Mandatory Live Web Research Before Reading Code

Eliminates stale training-cutoff analysis. Before inspecting the repo, the agent searches 2026 competitor acquisitions, shutdowns, funding rounds, CVEs, and regulatory shifts. Every matrix cell carries source URLs + verification dates, surfaced in the PDF Data Provenance Box.

"data_freshness": {
  "web_search_performed": true,
  "search_date": "2026-10-22",
  "queries_executed": 14,
  "sources_cited": 28
}
ZERO FALSE-POSITIVE NOISE

Code-Verified Red Flags Only (file:line + Actionable Fix)

In v0.6.2, heuristic false positives (git rebase timestamp diffs, high-velocity commits, documentation buzzword counts) are silently discarded and blocked by _filter_red_flags. Only genuine code-level security gaps, race conditions, and logic bugs with exact file:line and fix plans appear.

{
  "severity": "HIGH",
  "file_path": "src/crypto/session.rs:142",
  "title": "Nonce reuse risk on concurrent retry",
  "description": "Use atomic counter or ChaCha20Poly1305 random nonce..."
}
10-ITEM SITE & CODE AUDIT

10-Item Technical Audit (4 Claim-vs-Code + 6 Code-Measured)

Third-party badges (SOC2, ISO, HIPAA) are reference-only and never scored. DDE audits 4 claim-vs-code consistency checks plus 6 code-measured capabilities: Cryptographic depth, Concurrency model, I/O patterns, Caching architecture, Scalability, and ML implementation depth.

Claim vs Code (4): Pitch accuracy / Feature completeness
Code Measured (6): Crypto FFI / Concurrency / I/O / Cache / Scale / ML
왜 중요한가

AI 시대, 주장은 현실을 앞지른다

모든 피치덱이 "AI 기반"이라고 말합니다. 하지만 그것을 증명할 수 있는 곳은 드뭅니다. 기술 실사는 느리고, 비용이 많이 들며, 파이프라인 전반으로 확장하기 어렵습니다.

AI-washing은 어디에나 있다

마케팅 문구는 코드베이스가 뒷받침하지 못하는 기능을 과장하거나, 아예 지어냅니다. 그 간극을 짚어내려면 시니어 엔지니어의 시간이 필요합니다.

수작업 검토는 확장되지 않는다

딜마다 코드베이스를 읽고, 아키텍처를 파악하고, 경쟁사를 벤치마킹하는 일은 며칠에 걸친 전문가의 작업이며, 그럴 여유가 좀처럼 없습니다.

과장된 기대는 밸류에이션을 왜곡한다

객관적인 기술 기준선이 없으면, 가격과 리스크 판단은 증거 대신 창업자의 서사에 기대게 됩니다.

Dual Evaluation Architecture

5-Dimension Standard Scoring (20% Each) × Atlas 4-Axis Philosophy

Every report delivers both the balanced 5-Dimension Standard Score (equal 20% weights) and the parallel Atlas 4-Axis Engineering Philosophy (55% weighted toward deep cybersecurity engineering).

20%Lv.9/10

1. Technical Originality

API Wrapper vs. Genuine IP

Distinguishes thin third-party API wrappers from defensible proprietary algorithms, protocols, and custom engineering.

20%Lv.9/10

2. Technology Advancement

Stack Modernity & Cryptographic Edge

Evaluates 2026-grade language safety, runtime efficiency, post-quantum primitives (ML-KEM-1024), and modern toolchain maturity.

20%Lv.8/10

3. Implementation Depth

PoC Demo vs. Production Grade

Inspects error handling, edge cases, migrations, state machines, and automated test coverage to separate demos from production code.

20%Lv.8/10

4. Architecture Quality (incl. Security Posture)

Structural Integrity + Integrated Security (v0.3)

Evaluates security posture as an inseparable foundation of production architecture alongside modularity, fault isolation, and scalability.

20%Lv.8/10

5. Claim Consistency

Pitch Deck & Website Claims vs. Code Reality

Cross-examines public marketing claims ('AI-powered', 'E2EE', 'Zero-Knowledge') against actual repository implementation to expose AI-washing.

Grade Bands & Investment Recommendations

* In DDE analyze mode: Final Score = Heuristic Analysis (30%) + AI Average Score (70%), capped at 40 if any CRITICAL deal-breaker is present. Every dimension is rated 0–100 and Lv.1–10.
Grade A90–100

Strong investment candidate

Grade B75–89

Viable with conditions

Grade C60–74

Significant concerns

Grade D40–59

High risk

Grade F0–39

Do not invest (or Critical Deal-Breaker)

7 Competitive Charts × 6 Markets + 30-Item Matrix

Multi-Market Positioning & Sourced Implementation Capability Matrix

Generates 7 consulting frameworks across 6 global regions, plus an 8th chart: a ~30-item × 5–10 competitor Implementation Capability Matrix dynamically tailored to the target's industry with 3–5 line competitor selection rationales.

6 Global Markets (6–16 Competitors Plotted per Chart)
GlobalUSEMEAJapanSEALATAM

7 Consulting Frameworks (Each with Axis Rationale Captions)

Chart 01

Forrester Wave / Magic Quadrant

Vision × Execution

Chart 02

BCG Growth-Share Matrix

Market Growth × Relative Share

Chart 03

McKinsey Tech Moat

Competitive Position × Technical Moat Depth

Chart 04

Security & Privacy Maturity

Security Implementation × Privacy Readiness

Chart 05

Data Governance & Transparency

Data Protection × Audit Transparency

Chart 06

Goldman Sachs Risk-Adjusted Return

Downside Risk × Upside Potential

Chart 07

Innovation vs. Commercialization

R&D Velocity × ARR Traction (3D Bubble)

The 8th Chart: Implementation Capability Matrix (~30 Items × 5–10 Competitors)

Uses the 4-state Japanese tech rating standard (○ / △ / × / ?) with primary source URLs and verification dates on every cell.

○
verified

Publicly documented & verified in code, whitepapers, or specs

△
claimed

Asserted in marketing, but not independently verifiable

×
not_implemented

Explicitly absent in documentation or architecture

?
unknown

Cannot be determined from public evidence (preferred over guessing)

Category / CapabilityTargetComp. AComp. BComp. CProvenance (sources + last_verified)
Core Cryptography
Hybrid Post-Quantum Key Agreement (PQXDH / ML-KEM-1024)
○○×△[2026-10-22] Verified FIPS 203 ML-KEM-1024 + X25519 FFI implementation
Metadata Privacy
Sealed Sender / Zero-Knowledge Routing
○○××[2026-10-22] Verified sender certificate encapsulation across relays
Data Governance
Cryptographic Message Retirement (Tombstone Zeroize)
○△×?[2026-10-22] Verified Ed25519-signed Tombstone & cache zeroization
Audit Transparency
Public Source Code & Automated Security CI
○○××[2026-10-22] Verified public GitHub repository & CodeQL/osv-scanner CI

* In live PDF reports, the ~30 capability items and 5–10 top competitors are dynamically selected for the target's industry (Messaging, Fintech, Medical, Gaming, SaaS, IoT), with all cited URLs listed in the Source Appendix.

Inside the 24-Page Consulting PDF

Board- & Investment-Committee-Ready Deliverable

Typeset in black and Atlas Blue (#5271FF), combining executive synthesis, code-level evidence, global market charts, and a cryptographic purge certificate (~22–24 pages depending on findings).

P.01–03v0.6.3

Cover · Score Dashboard & Data Provenance Box · Executive Summary

5-dim horizontal bars, 5-dim radar chart, Data Provenance Box (search date, query & source counts, or training-cutoff warning), and business + technical synthesis.

P.04–09

SWOT 2×2 · Score Breakdown · Tech Level (Lv.1–10) · 1/3/5-Yr Outlook · Strategy · Thesis

Evidence-backed SWOT grid, per-dimension enablers, 1/3/5-year projections with confidence, immediate/medium/long-term roadmap, and investment thesis.

P.10–11v0.6.3

Code-Verified Red Flags (Zero False-Positive Noise) · 10-Item Site Verification

Genuine code defects only (file:line + fix plan) and the 10-item technical audit (4 claim-vs-code + 6 code-measured).

P.12–16v0.6.3

Competitive Landscape (7 Charts × 6 Markets) · Selection Rationales · 30-Item Matrix

Regional positioning charts, 3–5 line rationale per competitor (HQ, market position, why chosen), and the ○△×? capability matrix.

P.17–18

Atlas 4-Axis Dashboard · 5-Layer Security Sub-Breakdown · Non-Engineer Glossary

Performance / Stability / Lightweight / Security Strength bars plus a plain-language executive glossary (MFA, SOC2, libsignal, PQXDH).

P.19–24v0.6.3

Consistency · Rebuild Cost Estimate · Live Web Source Appendix · Purge Certificate

Engineering rebuild time/cost estimation, complete URL Source Appendix (added in v0.6.0), and cryptographic workspace purge certificate.

기능

DDE가 평가하는 것

반복 가능하고 증거를 우선하는 기술 평가 — 모든 딜에 동일한 평가 기준을 적용합니다.

주장 검증

피치 내용을 실제 코드베이스와 교차 검증하여, 실제로 구현된 것과 단지 주장에 그친 것을 가려냅니다.

AI-washing 탐지

"AI"가 얇은 래퍼에 불과한지, 하드코딩된 휴리스틱인지, 아예 존재하지 않는지를 짚어내어 — 진짜 머신러닝과 마케팅을 구분합니다.

신뢰도 기반 점수 산정

동등하게 가중된 항목 전반에 걸쳐 0–100 점수와 등급을 산출하며, 그 판단 근거를 함께 보여줍니다.

코드베이스 및 아키텍처 검토

구조, 의존성, 설계 선택을 정리해 유지보수성, 확장성, 기술 부채를 평가합니다.

실시간 경쟁 구도

실제 경쟁사, 펀딩, 보안 권고를 웹에서 조사하여 출처가 명시된 시장 비교 자료를 구성합니다.

설계부터 로컬·프라이빗

전적으로 IDE의 AI 샌드박스 안에서 동작합니다. 외부 서버도, 텔레메트리도 없으며 — CodeQL, Dependabot, pip-audit으로 감사됩니다.

Data Boundary & v0.6.3 (780e3cd / aac4c77) Hardening

Where Does Your Code Go? — 3 Execution Modes & Defense-in-Depth

DDE itself operates zero third-party servers and zero telemetry. Choose the exact execution boundary that matches your security policy.

Recommended · Zero API Keys

Mode 1: IDE AI Agent Execution (Standard)

$ dde prompt --pdf

DDE locally generates a structured prompt. Code reading and PDF generation happen inside your existing IDE AI agent session (Claude Code, Codex, Gemini CLI, Antigravity, Copilot, Cursor).

DDE makes zero outbound calls. Your AI agent runs STEP 0 live web searches for public competitor & CVE data.

$0 DDE fee. Uses your existing IDE AI subscription allowance.
100% Offline · Air-Gapped

Mode 2: Air-Gapped Local Heuristics (--skip-ai)

$ dde analyze . --skip-ai

Source code is analyzed 100% locally in memory/temp workspace and never sent to any AI model or external service.

In v0.6.3 (780e3cd / aac4c77), --skip-ai strictly disables API key auto-loading and blocks all outbound network requests.

100% free ($0, no API keys, no network connection required).
Optional · Multi-AI BYOK

Mode 3: Direct BYOK Multi-AI Cross-Verification ([byok] extra)

$ dde analyze owner/repo

Uses your own API keys to cross-verify across up to 3 providers via the unified SSOT registry (BYOK_DEFAULT_MODELS / PROVIDER_MODEL_COSTS): Anthropic (Claude Haiku 5.5 → Sonnet 5.5 → Fable 5.1), Google (Gemini 4 Argon), and OpenAI (Astra gpt-6.1-sol / GPT-5).

Configured for zero-retention API endpoints (store=False, with fail-closed enforcement on GoogleProvider). Strips user:token@ credentials from GitHub URLs and immediately dereferences PATs after single in-memory use.

$0 DDE fee (standard API token usage billed by your AI providers).
v0.6.3 · Commit 780e3cd · 156 Automated Tests

v0.6.3 (Commit 780e3cd / aac4c77) Security & Zero-Drift SSOT Hardening

Eradicated all 36 audit defects across 35+ modules with GPT-6-Astra (xhigh) & Gemini 4 Argon, unified the SSOT model/cost registry, and locked permanent verification with 156 automated tests (tests/test_audit_regressions.py + tests/test_zero_drift_invariants.py).

ENCRYPTED INGEST & SHREDsecure_loader.py · secure_delete.py

Secure Workspace, Zip Slip/FIFO Blocking & In-Place 3-Pass Shredding

Blocks Zip Slip traversal, symlinks, and FIFO special files during archive ingestion, and strips user:token@ credentials from GitHub URLs. Preserves .git during analysis for commit forensics, then performs in-place "r+b" 3-pass zero-fill shredding with strict os.walk(onerror=...) failure propagation and Windows NTFS user-only ACL locking.

SSRF & DNS REBINDING PINNINGsite_analyzer.py · _pinned_dns_context

ip.is_global + DNS-Pinned SSRF, CGNAT & TOCTOU Rebinding Defense

Site verification (site_analyzer.py) enforces ip.is_global to block RFC 6598 CGNAT (100.64.0.0/10), private, loopback, and cloud metadata IPs as well as unsafe redirects, using reentrant DNS pinning (_pinned_dns_context) to prevent time-of-check to time-of-use (TOCTOU) DNS rebinding attacks.

XML ESCAPE & 14-LANG SHAPINGpdf_generator.py · scorer.py · response_parser.py

ReportLab XML Escape & Arabic/Thai Helvetica Alphanumeric Protection

Applies _esc_text across all dynamic PDF fields (ImplementationMatrix, sources, RedFlags) to prevent ReportLab XML parser crashes. Protects Latin digits and ASCII punctuation with Helvetica in Arabic (ar) and Thai (th) to prevent missing-glyph tofu boxes (□) across 24-page PDFs, enforces Jinja2 autoescape=True, and guarantees idempotent Score.compute().

ZERO-DRIFT SSOT & 2026 AI SDKconfig.py · providers.py · test_zero_drift_invariants.py

Unified SSOT Model/Pricing Registry & Zero-Drift CI Invariant Gate

Unifies __version__, BYOK_DEFAULT_MODELS, and PROVIDER_MODEL_COSTS (Astra gpt-6.1-sol $2/$10, gpt-6-astra $5/$25, gpt-5 $1.25/$10, Haiku >100k tiered rate) in config.py SSOT. Enforces fail-closed store=False on GoogleProvider, supports async def AST analysis, and locks 156 CI tests via test_zero_drift_invariants.py.

작동 방식

한 번의 설치에서 24페이지 리포트까지

DDE는 파일을 읽고 셸을 실행할 수 있는 모든 AI 에이전트가 수행할 수 있는 구조화된 프롬프트를 생성합니다. DDE 전용 플러그인이나 로그인은 필요 없습니다.

1 · 한 번만 설치

단 하나의 pip 명령으로 공개 저장소에서 설치하세요. 순수 로컬 Python — 계정도, 설정도 필요 없습니다.

2 · IDE에서 실행

대상 프로젝트에서 AI IDE 터미널에 DDE 명령을 실행하세요. 분석은 이미 구독 중인 AI가 수행합니다 — 추가 비용 없이.

3 · 분석 및 조사

에이전트가 수백 개의 파일을 읽고, 가중 항목 전반에 걸쳐 평가하며, 경쟁사·펀딩·보안 권고를 웹에서 실시간으로 조사합니다.

4 · 점수화된 PDF 받기

컨설팅 스타일의 PDF가 Downloads 폴더에 도착합니다 — 0–100 점수, 등급, 경쟁 구도, 그리고 출처가 명시된 근거까지.

1-Step Quick Start & 14-Language PDF

Copy-Paste into Your AI Agent or Run One Line in Terminal

New in v0.6.3: paste Method A directly into your AI agent chat for fully automated install + audit + PDF generation, or run Method B in your terminal. Bundled Noto fonts render 14 PDF languages out of the box.

✓ Your current locale is natively supported for localized PDF output (14 languages).

Method A: Paste directly into your AI agent chat (Claude Code / Codex / Gemini CLI / Antigravity)
Install https://github.com/Atlas-Associates-Inc/Due-diligence-engine (python3 -m pip install -U --no-cache-dir git+https://github.com/Atlas-Associates-Inc/Due-diligence-engine.git), run dde prompt --pdf --lang ko on this project, and execute the entire generated prompt (STEP 0 live web research, full code audit, JSON generation, and PDF rendering) end-to-end to save the consulting PDF report to ~/Downloads/.
Method B: Paste 1 line in terminal (install + prompt generation in one step)
python3 -m pip install -U --no-cache-dir git+https://github.com/Atlas-Associates-Inc/Due-diligence-engine.git && dde prompt --pdf --lang ko
Verify environment (git, bundled fonts, reportlab, ~/Downloads, AI terminal)
dde doctor
Optional: Non-interactive URL verification
dde prompt --pdf --lang ko --url https://example.com
Verified AI Terminals & Models (v0.6.3 Tested)

Works out of the box in any AI agent capable of reading files and running shell commands.

Claude Code (Fable 5.1 / Sonnet 5.5 / Haiku 5.5)OpenAI Codex CLI (Astra GPT-6.1-Sol / GPT-5)Google Gemini CLI & Antigravity (Gemini 4 Argon)GitHub Copilot Agent ModeCursor / Windsurf / Aider / Continue.dev
vs. Other Tools

DDE vs. Manual DD vs. Generic AI Code Review vs. SaaS DD Platforms

Boutique consulting depth with local-first privacy at zero added software cost.

DimensionDDE (v0.6.3)Manual DD ConsultingGeneric AI Code ReviewSaaS DD Platforms
Added Cost$0 (uses your IDE AI)$$$$ (consultant fees)Per-token API fees$$$$ (enterprise subscription)
Code PrivacyLocal / IDE sandbox (0 DDE servers)Manual NDA & repo sharingUploads code to vendorStores code in vendor cloud
Deliverable24-page consulting PDF (14 langs)Bespoke reportInline PR commentsWeb dashboard
Data FreshnessLive web search (2026) + URLsDepends on analystTraining cutoff only (stale)Vendor refresh cycle
Crypto & Security DepthPQXDH / ML-KEM-1024 / FFI auditDepends on consultantSurface checklistSurface checklist
Competitive Benchmarks7 charts × 6 markets + 30-item matrixManual research (weeks)NoneLimited
Red Flag PrecisionCode-verified only (file:line + fix)High (slow)High false-positive noiseHeuristic rule noise
Setup & Turnaround1 command / ~10–20 min2–4 weeksMinutes (partial)Days (onboarding)
누구를 위한 것인가

진짜 기술력을 가려내야 하는 모든 이를 위해

투자 판단을 내리든, 자사 엔지니어링을 객관적으로 들여다보든, 경쟁사를 벤치마킹하든 — 코드에서 나온 증거를, 빠르게.

투자자 및 실사 담당

VC, 그로스 펀드, M&A 팀: 심층 분석에 들어가기 전, 파이프라인 전반의 기술 리스크와 주장 이면의 진실을 선별하세요.

자사 기술을 객관적으로 평가하려는 기업 및 CTO

자사 코드베이스에 DDE를 실행해 제3자의 시선으로 강점, 약점, 기술 부채를 확인하세요 — 개선 우선순위를 정하거나 다음 라운드를 준비하기 위해.

경쟁사를 벤치마킹하는 팀

공개 저장소와 시장 데이터를 바탕으로 경쟁사의 기술 스택과 구현 깊이를 벤치마킹하고, 증거를 통해 자사의 위치를 파악하세요.

FAQ

자주 묻는 질문

Due Diligence Engine은 무료인가요?

네. DDE는 Apache 2.0 라이선스의 오픈소스입니다. IDE에 이미 가지고 있는 AI 구독을 사용하므로 API 키도, 추가 비용도 없습니다.

제 코드가 기기 밖으로 나가나요?

아니요. DDE는 100% 로컬 Python이며 IDE의 AI 샌드박스 안에서 동작합니다. DDE 서버도, 텔레메트리도 없습니다. 웹 조사는 경쟁사와 시장에 관한 공개 정보만 가져옵니다.

어떤 AI 도구로 DDE를 실행할 수 있나요?

파일을 읽고 셸 명령을 실행할 수 있는 에이전트라면 무엇이든 가능합니다. Claude Code, OpenAI Codex CLI, Google Gemini CLI, Antigravity, GitHub Copilot Agent Mode에서 검증되었으며(Claude Opus 5.5 / Sonnet 5.5, Gemini 4 Argon, GPT-5 등 최신 모델 지원), Cursor, Gemini Code Assist, Aider, Windsurf 등 유사한 에이전트형 터ミ널에서도 동작합니다.

결과물은 어떤 형태인가요?

Downloads 폴더에 저장되는 컨설팅 수준의 PDF(약 24페이지)로, 전체 0–100 점수, 등급, 경쟁 구도, 구현 매트릭스, 그리고 출처가 명시된 근거를 담고 있습니다.

"AI-washing"이란 무엇이며 DDE는 이를 어떻게 탐지하나요?

AI-washing이란 기반 기술이 미미하거나 존재하지 않는데도 제품을 AI 기반이라고 마케팅하는 것을 말합니다. DDE는 피치에 담긴 주장을 코드베이스의 실제 구현과 비교하여 그 간극을 짚어냅니다.

오늘 첫 리포트를 실행하세요

한 번 설치하고, 어떤 AI IDE에서든 실행하여, 증거를 직접 확인하세요. 무료이자 오픈소스입니다.