🛡️ ForgeGuard

Autonomous AI Code Vulnerability Scanner, STRIDE Threat Modeler & Patch Engine

🎥 Watch the 3.5-Minute Full Demo Walkthrough: https://youtu.be/QicTSnnBr9k

ForgeGuard is an enterprise-grade cybersecurity platform that unites static abstract syntax tree (AST) vulnerability auditing, automated STRIDE threat modeling, and autonomous AST-verified code remediation into a unified tri-agent AI system.

ForgeGuard operates with zero heavy external dependencies (pure Python standard library core), achieves sub-10ms scan speeds, and guarantees zero broken builds via strict pre-commit AST verification and 1-click snapshot rollback safety.

---

🌟 Key Capabilities

1. Sub-10ms SAST AST Scanner:
   • Analyzes Python source trees via recursive AST visitors and dynamic taint-flow tracking.
   • Detects OWASP Top 10 / CWE vulnerabilities (SQL Injection, Command Injection, Insecure Deserialization, Hardcoded Secrets, Path Traversal, Weak Crypto, Predictable PRNGs, Eval Injection).
2. Automated STRIDE Threat Engine:
   • Maps detected entry points to all 6 STRIDE categories: Spoofing, Tampering, Repudiation, Information Disclosure, Denial of Service, and Elevation of Privilege.
   • Computes asset blast radius, attack surfaces, and architectural mitigations in real-time.
3. Self-Healing Autonomous Patcher:
   • Synthesizes standardized unified diffs with secure modern APIs (parameterized queries, subprocess.run(shell=False), json.loads, os.environ.get, secrets).
   • Validates patched code against Python AST before writing to disk.
   • Creates automatic atomic backups in .forgeguard_backups/ with instant rollback capabilities.
4. Featherless AI Inference Bridge:
   • Connects to open-source foundation models (Meta-Llama-3.1-8B-Instruct) on Featherless.ai for contextual threat intelligence with a deterministic offline heuristic synthesizer fallback.
5. Interactive Cyberpunk Web Dashboard:
   • Real-time Tailwind CSS dark-mode UI with CVSS 3.1 gauges, STRIDE matrix inspector, side-by-side diff studio, and 1-click auto-patch triggers.

---

🏗️ Tri-Agent Collaborative Architecture

flowchart TD
    subgraph Input ["Target Codebase / PR"]
        SourceCode["Python Source Files & Configs"]
    end

    subgraph TriAgent ["ForgeGuard Tri-Agent Engine"]
        Agent1["Agent 1: Security Auditor\n(AST SAST + Taint Flow + CVSS 3.1)"]
        Agent2["Agent 2: Threat Modeler\n(STRIDE Framework + Attack Blast Radius)"]
        Agent3["Agent 3: Remediation Engineer\n(Unified Diff + AST Syntax Gate)"]
    end

    subgraph SafetyGate ["Safety & Verification Engine"]
        ASTCheck{"AST Syntax Valid?"}
        Backup["Create Atomic Snapshot in .forgeguard_backups/"]
        DiskCommit["Apply Patch to Disk"]
        Abort["Abort Patch & Rollback"]
    end

    subgraph Output ["Deliverables & Interfaces"]
        CLI["Rich Terminal CLI"]
        WebUI["Cyberpunk Live Dashboard (Port 8899)"]
        JSONReport["Executive JSON / PDF Report"]
    end

    SourceCode --> Agent1
    Agent1 --> Agent2
    Agent1 --> Agent3
    Agent2 --> JSONReport
    Agent3 --> ASTCheck
    ASTCheck -- Yes --> Backup --> DiskCommit --> WebUI
    ASTCheck -- No --> Abort
    Agent1 --> CLI
    Agent2 --> WebUI

Loading

---

🚀 Quickstart Guide

1. Installation

Clone the repository and run ForgeGuard directly using Python 3.11+:

git clone https://github.com/kxrn-gupta/forgeguard.git
cd forgeguard

Zero pip installs required for core engine.

2. Run High-Speed SAST Scan

Scan any file or project directory:

python -m forgeguard.cli scan examples/vulnerable_app/app.py

3. Generate Full Executive Audit with STRIDE Threat Model

python -m forgeguard.cli audit examples/vulnerable_app/ --json audit_report.json

4. Autonomous Dry-Run or Real Code Patching

# Preview patches without modifying code
python -m forgeguard.cli patch examples/vulnerable_app/ --dry-run

# Safely apply all verified patches to disk
python -m forgeguard.cli patch examples/vulnerable_app/

5. Launch Interactive Live Dashboard

python -m forgeguard.cli serve examples/vulnerable_app/ --port 8899

Open http://localhost:8899 in your browser to inspect findings, explore STRIDE matrices, and test 1-click remediation.

---

📋 Standardized Rules Catalog

Rule ID	Title	OWASP Category	CWE ID	Default CVSS
FG-SEC-001	SQL Injection via Formatted Query	A03:2021-Injection	CWE-89	9.8 (CRITICAL)
FG-SEC-002	OS Command Injection (shell=True)	A03:2021-Injection	CWE-78	9.8 (CRITICAL)
FG-SEC-003	Insecure Object Deserialization (pickle)	A08:2021-Integrity Failures	CWE-502	9.8 (CRITICAL)
FG-SEC-004	Hardcoded Credentials / API Secrets	A07:2021-Auth Failures	CWE-798	7.5 (HIGH)
FG-SEC-005	Arbitrary Path Traversal	A01:2021-Broken Access	CWE-22	8.6 (HIGH)
FG-SEC-006	Weak Cryptographic Hash (MD5 / SHA1)	A02:2021-Crypto Failures	CWE-327	5.9 (MEDIUM)
FG-SEC-007	Insecure PRNG for Security Tokens	A02:2021-Crypto Failures	CWE-338	5.3 (MEDIUM)
FG-SEC-008	Dynamic Code Evaluation (eval/exec)	A03:2021-Injection	CWE-95	9.8 (CRITICAL)
FG-SEC-009	Permissive CORS Wildcard Policy	A01:2021-Broken Access	CWE-942	3.7 (LOW)

---

🧪 Comprehensive Test Suite

ForgeGuard includes a full unit and integration test suite:

python -m unittest discover -s tests -p "test_*.py" -v

Results: 6/6 tests passing (0.044s) covering AST visitors, STRIDE modelers, patchers, rollback snapshots, and full engine reports.

---

🏆 Hackathon Metadata

• Event: ForgeHacks 2026 (Devpost)
• Track: AI + Cybersecurity
• Project Name: ForgeGuard (Autonomous AI Code Vulnerability Scanner, STRIDE Threat Modeler & Patch Engine)
• Slide Deck: PRESENTATION_SLIDES.pdf (8 slides, high-res PDF in repo root)
• License: MIT