Detection and hunting framework for Broken Object Level Authorization vulnerabilities in APIs.

Install

mkdir -p .claude/skills/bola-idor && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/11013" && unzip -o skill.zip -d .claude/skills/bola-idor && rm skill.zip

Installs to .claude/skills/bola-idor

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Use when hunting Broken Object Level Authorization (BOLA) or Insecure Direct Object Reference (IDOR) vulnerabilities in APIs or web applications. Trigger on: "BOLA", "IDOR", "broken object level", "access other users", "object reference", numeric or UUID IDs in URLs or request bodies, user-scoped resources, horizontal privilege escalation, "change the ID in the request", second-order IDOR, blind IDOR, indirect reference, encoded ID, deprecated API version, JSON globbing.
475 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Advanced

Key capabilities

  • Scan for IDOR vulnerabilities
  • Test object-level authorization
  • Fuzz numeric identifiers
  • Bypass access control

How it works

It tests if the server verifies object ownership before granting access to resources.

Inputs & outputs

You give it
API endpoint
You get back
Vulnerability assessment

When to use bola-idor

  • Hunting idor vulnerabilities
  • Testing api access control
  • Checking object level authorization

About this skill

Broken Object Level Authorization (BOLA / IDOR)

What Is Broken and Why

The server accepts a resource identifier from the client and fetches the object without verifying the requesting user owns or has access to it. Authorization is enforced at the route level ("is this user logged in?") but not at the object level ("does this user own object 1042?"). An attacker substitutes their identifier for a victim's to read, modify, or delete resources they should never access. BOLA is consistently the #1 OWASP API Security risk because it is trivial to test and almost always yields high-severity findings.

Key Signals

  • Numeric or sequential IDs in URL path: /api/orders/1042, /users/7/profile
  • UUIDs or hashes in query params or body referencing another user's object
  • Parameters named user_id, account_id, owner_id, ref, target_id, invoice
  • Write operations (PUT/PATCH/DELETE) accepting an object ID
  • Export/download/share endpoints with a resource ID parameter
  • Keywords me or current used as ID aliases — swappable for integer IDs
  • Older API versions still accessible: /v1/, /v2/, /legacy/
  • UUIDs discoverable via public profiles, share links, password reset flows, Wayback Machine

Methodology

  1. Map all object identifiers across the entire app — URLs, query params, request body, cookies, headers.
  2. Create two accounts with distinctive names (e.g. attacker_a, victim_b). Use separate browsers to keep sessions fully isolated. Fully populate the victim account with varied resources and document all encountered IDs. Capture all requests with the victim session.
  3. Replay each request using the attacker token with the victim's object IDs.
  4. Compare responses — same data returned = BOLA confirmed.
  5. Test unauthenticated: remove Authorization header entirely.
  6. Test write operations (PUT/PATCH/DELETE) — impact is higher than reads.
  7. Test indirect references: export endpoints, share links, scheduled jobs, email triggers.
  8. Test less-visible features: auto-save, draft, notification, audit log, attachment endpoints.
  9. Try older API versions (/v1/, /beta/) which often lack access control patches.
  10. For second-order IDOR: store a payload referencing victim's ID, trigger async processing, observe outcome.

Payloads & Tools

# ffuf: fuzz numeric IDs around your own
ffuf -w <(seq 1000 2000) -u https://TARGET/api/users/FUZZ \
  -H "Authorization: Bearer YOUR_TOKEN" -mc 200 -fs 0

# curl: direct ID swap
curl -s https://TARGET/api/orders/VICTIM_ID \
  -H "Authorization: Bearer YOUR_TOKEN"

# Append .json to bypass access control
curl https://TARGET/api/receipts/VICTIM_ID.json \
  -H "Authorization: Bearer YOUR_TOKEN"

# Try deprecated API version
curl https://TARGET/v1/users/VICTIM_ID \
  -H "Authorization: Bearer YOUR_TOKEN"
# Burp Intruder: fuzz ±1000 around your own ID
GET /api/orders/§1042§ HTTP/1.1
Authorization: Bearer YOUR_TOKEN

# JSON globbing in request body
{"user_id": [YOUR_ID, VICTIM_ID]}
{"user_id": "*"}
{"user_id": true}
{"user_id": 0}
{"user_id": -1}
{"user_id": 1235.0}

Burp extension: Autorize — automatically replaces session token with low-priv token on every request, flags unexpected 200s and response diffs.

Bypass Techniques

  • JSON globbing: replace ID with [id1, id2], *, true, 0, -1, 1234.0 — parsers may match all
  • HTTP verb swap: GET blocked → try POST, PUT, DELETE, PATCH on same path
  • Parameter pollution: ?user_id=YOURS&user_id=VICTIM — server may process last or first
  • Encoded references: base64/hex decode the ID, increment, re-encode — app trusts encoding as security
  • Alternate field names: owner_id, account_id, ref, target, resource_id, parent_id
  • Path traversal: /api/users/YOURS/../VICTIM
  • Static keyword swap: replace me or current with a numeric ID
  • Content-type switch: JSON endpoint may behave differently with application/x-www-form-urlencoded
  • Second-order: store a reference to victim's ID in a field, trigger async job that processes it without re-checking auth
  • Append extension: /resource/VICTIM_ID.json, .xml, .csv may skip access control middleware

Exploitation Scenarios

Scenario 1 — Account takeover via email change Setup: PUT /api/users/{id} accepts email as an editable field, no ownership check. Trigger: Attacker replaces their own id with victim's id in the request body. Impact: Victim's email changed to attacker's address → password reset → full account takeover.

Scenario 2 — Mass PII leak via sequential ID Setup: /api/orders/{id} returns full order: name, address, card last4, phone. Trigger: Attacker iterates integer IDs from 1 to N with their own session token. Impact: Thousands of customers' PII and payment metadata exfiltrated via scripted enumeration.

Scenario 3 — Second-order IDOR via scheduled export Setup: App lets users schedule data exports; export job runs async and emails result. Trigger: Attacker sets export_for_user_id=VICTIM_ID in the schedule request. Impact: Victim's full data export emailed to attacker — no access control on the async job.

False Positives

  • API returns 200 but with empty or redacted data — access control is working, just silent
  • Public resources (product listings, public profiles) — no authorization expected by design
  • Response identical regardless of ID — server reads from session context, ID param is ignored
  • me and current aliases that correctly resolve to the authenticated user only

Fix Patterns

-- Correct: ownership enforced at query level
SELECT * FROM orders WHERE id = ? AND user_id = current_user_id()
# Wrong: fetch then check (inefficient + race-prone)
order = db.find(id)
if order.user_id != current_user:
    raise Forbidden()

# Correct: indirect reference map (never expose raw DB IDs)
user_resource_map = {session_token: [allowed_id_1, allowed_id_2]}
if requested_id not in user_resource_map[session_token]:
    raise Forbidden()
  • Use indirect reference maps: expose opaque tokens that map server-side to real IDs
  • Centralize authorization in middleware — one place, not scattered per-route
  • Apply access control consistently across ALL HTTP methods and API versions
  • Disable or equally secure deprecated API versions

Related Skills

[[authz-bypass]] covers the broader authorization failure class — BOLA is its most common manifestation. When the application uses GraphQL, [[graphql-idor-via-introspection-leak]] shows how to enumerate the schema to find every object type accepting an ID argument. [[path-traversal]] is an IDOR on the filesystem: the same "reference to a resource without ownership check" pattern applied to file paths. IDOR findings frequently reveal [[business-logic-flaws]] — such as skipping payment by referencing another order's paid state.

When not to use it

  • General security scanning

Prerequisites

Two test accounts

Limitations

  • Requires manual setup of test accounts
  • High false positive rate

How it compares

It focuses specifically on object-level authorization failures rather than general access control.

Compared to similar skills

bola-idor side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
bola-idor (this skill)02moReviewAdvanced
reverse-engineering-tools734moNo flagsAdvanced
game-hacking-techniques422moNo flagsAdvanced
solidity-security152moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

You might also like

reverse-engineering-tools

gmh5225

Guide for reverse engineering tools and techniques used in game security research. Use this skill when working with debuggers, disassemblers, memory analysis tools, binary analysis, or decompilers for game security research.

73204

game-hacking-techniques

gmh5225

Guide for game hacking techniques and cheat development. Use this skill when researching memory manipulation, code injection, ESP/aimbot development, overlay rendering, or game exploitation methodologies.

42128

solidity-security

wshobson

Master smart contract security best practices to prevent common vulnerabilities and implement secure Solidity patterns. Use when writing smart contracts, auditing existing contracts, or implementing security measures for blockchain applications.

15115

1password

openclaw

Set up and use 1Password CLI (op). Use when installing the CLI, enabling desktop app integration, signing in (single or multi-account), or reading/injecting/running secrets via op.

2799

senior-security

davila7

Comprehensive security engineering skill for application security, penetration testing, security architecture, and compliance auditing. Includes security assessment tools, threat modeling, crypto implementation, and security automation. Use when designing security architecture, conducting penetration tests, implementing cryptography, or performing security audits.

3191

ghidra

mitsuhiko

Reverse engineer binaries using Ghidra's headless analyzer. Decompile executables, extract functions, strings, symbols, and analyze call graphs without GUI.

16105

Search skills

Search the agent skills registry