File size: 16,169 Bytes
8df6aa0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d38f080
8df6aa0
 
 
 
d38f080
8df6aa0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d38f080
8df6aa0
 
 
 
 
d38f080
 
8df6aa0
 
 
 
 
d38f080
 
8df6aa0
 
 
 
 
 
 
 
 
 
 
 
 
 
d38f080
8df6aa0
d38f080
8df6aa0
d38f080
8df6aa0
d38f080
8df6aa0
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
#!/usr/bin/env python3
"""AppSecBench vulnerability catalog.

Each entry is an ORIGINAL archetype definition: the canonical CWE, the OWASP
Top 10 (2021) and OWASP API Security (2023) mapping, the OWASP LLM Top 10
(2025) mapping where relevant, the high-level category, representative public
references (OWASP/CWE/PortSwigger/OWASP ASVS — no copyrighted code), and a
human-readable description template that the generator fills in per case.

Nothing here is copied from an existing benchmark; the code snippets are
authored from scratch in `generators.py`.
"""
from __future__ import annotations

# Canonical list of supported dimensions (derived from the master prompt).
LANGUAGES = [
    "Python", "Java", "JavaScript", "TypeScript", "Go", "Rust",
    "PHP", "C#", "Kotlin", "Swift", "C", "C++",
    "Ruby", "Scala",
]
FRAMEWORKS = [
    "Flask", "FastAPI", "Django", "Spring Boot", "Express", "NestJS",
    "Next.js", "Laravel", "ASP.NET Core", "Gin", "Echo", "Fiber",
    "Android", "iOS", "Rails", "Angular", "Vue",
]
DIFFICULTIES = ["Beginner", "Intermediate", "Advanced", "Expert", "Real-world enterprise"]
# Frameworks are only relevant for vulnerabilities that live in a web/rpc layer.
DIFFICULTIES_NO_FRAMEWORK = ["Beginner", "Intermediate", "Advanced", "Expert", "Real-world enterprise"]

# (vulnerability_name, category, cwe, owasp2021, owasp_api2023, owasp_llm2025, refs, description)
CATALOG = [
    ("SQL Injection", "Injection", "CWE-89", "A03:2021", "API8:2023", "",
     ["https://owasp.org/www-community/attacks/SQL_Injection",
      "https://cwe.mitre.org/data/definitions/89.html",
      "https://cheatsheetseries.owasp.org/cheatsheets/SQL_Injection_Prevention_Cheat_Sheet.html"],
     "Untrusted input is concatenated into a SQL query string, allowing an attacker to alter the query's structure and read or modify data."),
    ("Cross-Site Scripting", "Injection", "CWE-79", "A03:2021", "API8:2023", "",
     ["https://owasp.org/www-community/attacks/xss/",
      "https://cwe.mitre.org/data/definitions/79.html",
      "https://cheatsheetseries.owasp.org/cheatsheets/Cross_Site_Scripting_Prevention_Cheat_Sheet.html"],
     "Untrusted input is reflected into an HTTP response without output encoding, letting an attacker execute JavaScript in the victim's browser."),
    ("Server-Side Request Forgery", "Injection", "CWE-918", "A10:2021", "API7:2023", "",
     ["https://owasp.org/www-community/attacks/Server_Side_Request_Forgery",
      "https://cwe.mitre.org/data/definitions/918.html"],
     "A server fetches a URL built from user input, letting an attacker reach internal services, cloud metadata endpoints, or ports not exposed publicly."),
    ("Command Injection", "Injection", "CWE-77", "A03:2021", "API8:2023", "",
     ["https://owasp.org/www-community/attacks/Command_Injection",
      "https://cwe.mitre.org/data/definitions/77.html"],
     "User input is passed to a shell command interpreter, letting an attacker execute arbitrary operating-system commands."),
    ("Path Traversal", "Injection", "CWE-22", "A01:2021", "API8:2023", "",
     ["https://owasp.org/www-community/attacks/Path_Traversal",
      "https://cwe.mitre.org/data/definitions/22.html"],
     "User-controlled file paths are joined to a base directory without canonicalization, allowing read or write outside the intended root."),
    ("XML External Entity", "Injection", "CWE-611", "A05:2021", "API8:2023", "",
     ["https://owasp.org/www-community/vulnerabilities/XML_External_Entity_(XXE)_Processing",
      "https://cwe.mitre.org/data/definitions/611.html"],
     "An XML parser resolves external entities in attacker-supplied documents, enabling file disclosure, SSRF, or denial of service."),
    ("Insecure Direct Object Reference", "Broken Access Control", "CWE-639", "A01:2021", "API1:2023", "",
     ["https://owasp.org/www-community/attacks/Insecure_Direct_Object_Reference",
      "https://cwe.mitre.org/data/definitions/639.html"],
     "An endpoint exposes an object reference (id, filename, token) and trusts the caller to only access records they own, enabling horizontal privilege escalation."),
    ("Broken Access Control", "Broken Access Control", "CWE-285", "A01:2021", "API5:2023", "",
     ["https://owasp.org/Top10/A01_2021-Broken_Access_Control/",
      "https://cwe.mitre.org/data/definitions/285.html"],
     "Authorization checks are missing or can be bypassed, letting a user act on resources or functions outside their privilege level."),
    ("Broken Authentication", "Authentication", "CWE-287", "A07:2021", "API2:2023", "",
     ["https://owasp.org/Top10/A07_2021-Identification_and_Authentication_Failures/",
      "https://cwe.mitre.org/data/definitions/287.html"],
     "Authentication mechanisms are weak, spoofable, or bypassable, allowing an attacker to assume another user's identity."),
    ("Broken Authorization", "Authorization", "CWE-862", "A01:2021", "API5:2023", "",
     ["https://cwe.mitre.org/data/definitions/862.html",
      "https://owasp.org/Top10/A01_2021-Broken_Access_Control/"],
     "A function performs a sensitive action without verifying the caller is permitted to do so (missing authorization check)."),
    ("JWT Vulnerabilities", "Authentication", "CWE-345", "A07:2021", "API2:2023", "",
     ["https://cwe.mitre.org/data/definitions/345.html",
      "https://cheatsheetseries.owasp.org/cheatsheets/JSON_Web_Token_for_Java_Cheat_Sheet.html"],
     "A JWT implementation accepts tokens with a weak or 'none' algorithm, skips signature verification, or trusts an attacker-controlled key/secret."),
    ("OAuth Vulnerabilities", "Authentication", "CWE-287", "A07:2021", "API2:2023", "",
     ["https://cwe.mitre.org/data/definitions/287.html",
      "https://owasp.org/Top10/A07_2021-Identification_and_Authentication_Failures/"],
     "An OAuth/OIDC flow is misimplemented (redirect_uri not validated, implicit flow, token leakage) allowing account takeover."),
    ("Session Management", "Authentication", "CWE-614", "A07:2021", "API2:2023", "",
     ["https://cwe.mitre.org/data/definitions/614.html",
      "https://cheatsheetseries.owasp.org/cheatsheets/Session_Management_Cheat_Sheet.html"],
     "Session identifiers are created, stored, or invalidated insecurely (predictable IDs, missing HttpOnly/Secure, no rotation on auth)."),
    ("Cross-Site Request Forgery", "Broken Access Control", "CWE-352", "A01:2021", "API8:2023", "",
     ["https://owasp.org/www-community/attacks/csrf",
      "https://cwe.mitre.org/data/definitions/352.html"],
     "A state-changing request lacks an unguessable token, letting a third-party site force the victim's browser to perform actions."),
    ("Insecure File Upload", "Injection", "CWE-434", "A03:2021", "API8:2023", "",
     ["https://owasp.org/www-community/vulnerabilities/Unrestricted_File_Upload",
      "https://cwe.mitre.org/data/definitions/434.html"],
     "Uploaded files are not validated for type, content, or location, enabling remote code execution or storage abuse."),
    ("Insecure Deserialization", "Injection", "CWE-502", "A08:2021", "API8:2023", "",
     ["https://owasp.org/www-community/vulnerabilities/PHP_Object_Injection",
      "https://cwe.mitre.org/data/definitions/502.html"],
     "Untrusted data is deserialized by a native/object serializer, which can trigger code execution or logic abuse during reconstruction."),
    ("Open Redirect", "Injection", "CWE-601", "A01:2021", "API8:2023", "",
     ["https://cwe.mitre.org/data/definitions/601.html",
      "https://cheatsheetseries.owasp.org/cheatsheets/Unvalidated_Redirects_and_Forwards_Cheat_Sheet.html"],
     "A redirect target is taken from user input without an allow-list, enabling phishing by bouncing victims through a trusted domain."),
    ("Race Condition", "Business Logic", "CWE-362", "A04:2021", "", "",
     ["https://cwe.mitre.org/data/definitions/362.html",
      "https://owasp.org/www-community/attacks/Thread_Safety_and_State_Problems"],
     "A check-then-act sequence is not atomic, letting concurrent requests pass the check before a shared state update lands (TOCTOU)."),
    ("Insecure Cryptography", "Cryptographic Failures", "CWE-327", "A02:2021", "", "",
     ["https://cwe.mitre.org/data/definitions/327.html",
      "https://owasp.org/Top10/A02_2021-Cryptographic_Failures/"],
     "Broken or deprecated primitives (MD5/SHA1, ECB, static IVs, weak RNG) are used to protect sensitive data."),
    ("Hardcoded Secrets", "Cryptographic Failures", "CWE-798", "A02:2021", "", "",
     ["https://cwe.mitre.org/data/definitions/798.html",
      "https://owasp.org/Top10/A02_2021-Cryptographic_Failures/"],
     "Credentials, API keys, or private keys are embedded in source code, exposing them to anyone with repository access."),
    ("Business Logic", "Business Logic", "CWE-840", "A04:2021", "API6:2023", "",
     ["https://cwe.mitre.org/data/definitions/840.html",
      "https://owasp.org/Top10/A04_2021-Insecure_Design/"],
     "The application enforces no invariant for a workflow (negative quantity, replayed coupon, price override), leading to abuse."),
    ("Missing Rate Limiting", "Security Misconfiguration", "CWE-307", "A07:2021", "API4:2023", "",
     ["https://cwe.mitre.org/data/definitions/307.html",
      "https://owasp.org/Top10/A07_2021-Identification_and_Authentication_Failures/"],
     "Authentication or sensitive endpoints lack throttling, enabling credential stuffing, brute force, or resource exhaustion."),
    ("Sensitive Data Logging", "Security Misconfiguration", "CWE-532", "A09:2021", "", "",
     ["https://cwe.mitre.org/data/definitions/532.html",
      "https://owasp.org/Top10/A09_2021-Security_Logging_and_Monitoring_Failures/"],
     "Credentials, tokens, or PII are written to logs in cleartext, widening the blast radius of a log compromise."),
    ("Header Injection", "Injection", "CWE-113", "A03:2021", "API8:2023", "",
     ["https://cwe.mitre.org/data/definitions/113.html",
      "https://owasp.org/www-community/attacks/HTTP_Response_Splitting"],
     "Unsanitized input is placed into response or request headers, allowing response splitting or header smuggling."),
    ("Prompt Injection", "AI Security", "CWE-1427", "A05:2021", "", "LLM01:2025",
     ["https://cwe.mitre.org/data/definitions/1427.html",
      "https://owasp.org/www-project-top-10-for-large-language-model-applications/"],
     "An LLM application merges untrusted content into its prompt, letting an attacker override instructions or exfiltrate data."),
    ("RAG Security", "AI Security", "CWE-1427", "A05:2021", "", "LLM02:2025",
     ["https://owasp.org/www-project-top-10-for-large-language-model-applications/",
      "https://cwe.mitre.org/data/definitions/1427.html"],
     "A retrieval-augmented generation pipeline ingests untrusted documents whose content steers the model or poisons the index."),
    ("MCP Security", "AI Security", "CWE-918", "A05:2021", "", "LLM06:2025",
     ["https://modelcontextprotocol.io/",
      "https://owasp.org/www-project-top-10-for-large-language-model-applications/"],
     "A Model Context Protocol server exposes tools/resources that an agent can invoke with overly broad permissions or unvalidated arguments."),
    ("AI Agent Security", "AI Security", "CWE-77", "A05:2021", "", "LLM06:2025",
     ["https://owasp.org/www-project-top-10-for-large-language-model-applications/",
      "https://cwe.mitre.org/data/definitions/77.html"],
     "An autonomous agent that can call tools or shells is tricked by external content into performing unsafe actions."),
    ("GraphQL Security", "API Security", "CWE-915", "A04:2021", "API4:2023", "",
     ["https://cwe.mitre.org/data/definitions/915.html",
      "https://owasp.org/Top10/A04_2021-Insecure_Design/"],
     "A GraphQL endpoint lacks query costing/introspection control or object-level authorization, enabling data exfiltration or DoS."),
    ("REST API Security", "API Security", "CWE-639", "A01:2021", "API1:2023", "",
     ["https://owasp.org/Top10/A01_2021-Broken_Access_Control/",
      "https://cwe.mitre.org/data/definitions/639.html"],
     "A REST API exposes objects by id without ownership checks or proper input validation, enabling IDOR or mass assignment."),
    ("gRPC Security", "API Security", "CWE-285", "A01:2021", "API5:2023", "",
     ["https://cwe.mitre.org/data/definitions/285.html",
      "https://owasp.org/Top10/A01_2021-Broken_Access_Control/"],
     "A gRPC service does not enforce per-method authorization or input validation, allowing unauthorized RPC calls."),
    ("Cloud Misconfiguration", "Security Misconfiguration", "CWE-1188", "A05:2021", "", "",
     ["https://cwe.mitre.org/data/definitions/1188.html",
      "https://owasp.org/Top10/A05_2021-Security_Misconfiguration/"],
     "A cloud resource (bucket, blob, policy, security group) is exposed publicly or grants excessive permission."),
    ("Kubernetes Security", "Security Misconfiguration", "CWE-250", "A05:2021", "", "",
     ["https://cwe.mitre.org/data/definitions/250.html",
      "https://owasp.org/Top10/A05_2021-Security_Misconfiguration/"],
     "A pod or cluster is over-privileged (privileged container, hostPath mount, wildcard RBAC) enabling node compromise."),
    ("Docker Security", "Security Misconfiguration", "CWE-250", "A05:2021", "", "",
     ["https://cwe.mitre.org/data/definitions/250.html",
      "https://owasp.org/Top10/A05_2021-Security_Misconfiguration/"],
     "A container image or runtime is configured insecurely (running as root, mounting the Docker socket, latest tag, secrets in env)."),
]

# Build fast lookup
CATALOG_BY_NAME = {}
for _row in CATALOG:
    (_name, _cat, _cwe, _o21, _oapi, _ollm, _refs, _desc) = _row
    CATALOG_BY_NAME[_name] = {
        "category": _cat, "cwe": _cwe, "owasp": _o21,
        "owasp_api": _oapi, "owasp_llm": _ollm,
        "references": _refs, "description": _desc,
    }


# Which (vuln, language) pairs make sense. We restrict infra/AI vulns to a
# sensible subset so every record is realistic rather than contrived.
INFRA_VULNS = {"Cloud Misconfiguration", "Kubernetes Security", "Docker Security"}
AI_VULNS = {"Prompt Injection", "RAG Security", "MCP Security", "AI Agent Security"}
API_VULNS = {"GraphQL Security", "REST API Security", "gRPC Security"}

# Languages that trivially support a native web/rpc framework example.
WEB_LANGS = {"Python", "Java", "JavaScript", "TypeScript", "Go", "PHP", "C#", "Kotlin", "Swift", "Ruby", "Scala"}

# Framework applicability per language (kept realistic).
LANG_FRAMEWORKS = {
    "Python": ["Flask", "FastAPI", "Django"],
    "Java": ["Spring Boot"],
    "JavaScript": ["Express", "Next.js", "Angular", "Vue"],
    "TypeScript": ["Express", "NestJS", "Next.js", "Angular", "Vue"],
    "Go": ["Gin", "Echo", "Fiber"],
    "PHP": ["Laravel"],
    "C#": ["ASP.NET Core"],
    "Kotlin": ["Android", "Spring Boot"],
    "Swift": ["iOS"],
    "Ruby": ["Rails"],
    "Scala": ["Spring Boot"],
    "Rust": ["None"],   # no Rust framework in the required framework list
    "C": ["None"],
    "C++": ["None"],
}


def applicable_languages(vuln_name: str) -> list:
    if vuln_name in INFRA_VULNS:
        return ["YAML", "Dockerfile", "Bash"]
    if vuln_name == "Insecure Cryptography":
        return ["Python", "Java", "C", "C++", "Go", "JavaScript", "PHP"]
    if vuln_name == "Hardcoded Secrets":
        return LANGUAGES
    if vuln_name in AI_VULNS:
        return ["Python", "JavaScript", "TypeScript", "Ruby"]
    if vuln_name == "Race Condition":
        return ["Python", "Java", "Go", "JavaScript", "C", "C++", "PHP", "C#", "Ruby"]
    if vuln_name == "Insecure Deserialization":
        return ["Python", "Java", "PHP", "JavaScript", "C#", "Go", "Ruby"]
    if vuln_name in API_VULNS:
        return ["Python", "JavaScript", "TypeScript", "Go", "Java", "Ruby"]
    # Generic web/rpc injection & access-control classes.
    return [l for l in LANGUAGES if l in WEB_LANGS]


def applicable_frameworks(vuln_name: str, language: str) -> list:
    if vuln_name in INFRA_VULNS or language in ("YAML", "Dockerfile", "Bash", "C", "C++"):
        return ["None"]
    return LANG_FRAMEWORKS.get(language, ["None"])