| name | report-writing |
| description | Bug bounty report writing for H1/Bugcrowd/Intigriti/Immunefi — report templates, human tone guidelines, impact-first writing, CVSS 3.1 scoring, title formula, impact statement formula, severity decision guide, downgrade counters, pre-submit checklist. Use after validating a finding and before submitting. Never use "could potentially" — prove it or don't report. |
| sources | field_recon, hackerone_public, bugcrowd_public, immunefi_public |
| report_count | 50 |
REPORT WRITING
Impact-first. Human tone. No theoretical language. Triagers are people.
THE MOST IMPORTANT RULE
Never use "could potentially" or "could be used to" or "may allow".
Either it does the thing or it doesn't. If you haven't proved it, don't claim it.
BAD: "This vulnerability could potentially allow an attacker to access user data."
GOOD: "An attacker can access any user's order history by changing the user_id
parameter to the target user's ID. I confirmed this using two test accounts:
attacker@test.com (ID 123) successfully retrieved victim@test.com (ID 456)
orders, including their shipping address and payment method last 4 digits."
TITLE FORMULA
[Bug Class] in [Exact Endpoint/Feature] allows [attacker role] to [impact] [victim scope]
Good titles (specific, impact-first):
IDOR in /api/v2/invoices/{id} allows authenticated user to read any customer's invoice data
Missing auth on POST /api/admin/users allows unauthenticated attacker to create admin accounts
Stored XSS in profile bio field executes in admin panel — allows privilege escalation
SSRF via image import URL parameter reaches AWS EC2 metadata service
Race condition in coupon redemption allows same code to be used unlimited times
Bad titles (vague, useless to triager):
IDOR vulnerability found
Broken access control
XSS in user input
Security issue in API
Unauthorized access to user data
HACKERONE REPORT TEMPLATE
## Summary
[One paragraph: what the bug is, where it is, what an attacker can do. Be specific.
Include: endpoint, method, parameter, data exposed, required access level.]
Example: "The `/api/users/{user_id}/orders` endpoint does not verify that the
authenticated user owns the requested user_id. An attacker can enumerate any
user's order history, including PII (email, address, phone) and purchase history,
by incrementing the user_id parameter. No privileges beyond a standard free
account are required."
## Vulnerability Details
**Vulnerability Type:** IDOR / Broken Object Level Authorization
**CVSS 3.1 Score:** 6.5 (Medium) — AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N
**Affected Endpoint:** GET /api/users/{user_id}/orders
## Steps to Reproduce
**Environment:**
- Attacker account: attacker@test.com, user_id = 123
- Victim account: victim@test.com, user_id = 456
- Target: https://target.com
**Steps:**
1. Log in as attacker@test.com, obtain Bearer token
2. Send the following request:
GET /api/users/456/orders HTTP/1.1
Host: target.com
Authorization: Bearer ATTACKER_TOKEN_HERE
3. Observe response:
```json
{
"orders": [
{"id": 789, "items": [...], "email": "victim@test.com", "address": "123 Main St..."}
]
}
The response contains victim's full order history and PII despite being requested
by a different user.
Impact
An authenticated attacker can enumerate all user orders by iterating user_id values.
This exposes: full name, email, shipping address, purchase history, and payment
method (last 4). With ~100K users, this represents a mass PII breach affecting
all registered users. Exploitation requires only a free account and takes minutes
with a simple loop.
Recommended Fix
Add server-side ownership verification:
if order.user_id != current_user.id:
raise Forbidden()
Supporting Materials
[Screenshot showing attacker's session returning victim's order data]
[Video walkthrough if available]
---
## BUGCROWD REPORT TEMPLATE
```markdown
# [IDOR] User order history accessible without authorization via /api/users/{id}/orders
**VRT Category:** Broken Access Control > IDOR > P2
## Description
[Same impact-first paragraph as HackerOne summary]
## Steps to Reproduce
[Same structured steps — exact HTTP requests, exact responses]
## Proof of Concept
[Screenshot/video showing the actual impact]
## Expected vs Actual Behavior
**Expected:** 403 Forbidden when user_id does not match authenticated user
**Actual:** 200 OK with victim's full order data
## Severity Justification
P2 (High) — Direct read access to other users' PII. Affects all user accounts.
No user interaction required. Exploitable by any authenticated user.
Automated enumeration could exfil all [N] user records in minutes.
## Remediation
Add ownership verification: `if order.user_id != current_user.id: raise 403`
INTIGRITI REPORT TEMPLATE
# [Bug Class]: [Exact Impact] in [Endpoint/Feature]
## Description
[Impact-first paragraph. Start with what an attacker can do, not with how you found it.
Include: endpoint, method, parameter, data exposed, required privileges.]
## Steps to Reproduce
**Environment:**
- Attacker: email=attacker@test.com (standard account, no special role)
- Victim: email=victim@test.com
- Tested: [date]
**Reproduction steps:**
1. [Login as attacker / visit URL / send request]
2. Send the following HTTP request:
\```http
METHOD /endpoint HTTP/1.1
Host: target.com
Authorization: Bearer ATTACKER_TOKEN
Content-Type: application/json
{"param": "victim_id_here"}
\```
3. Observe response contains victim's private data:
\```json
{"email": "victim@test.com", "address": "123 Main St", ...}
\```
## Impact
[Specific, quantified impact. What data, how many users, what can attacker do.]
CVSS 3.1 Score: X.X ([Severity]) — AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N
## Remediation
[1-3 sentence concrete fix. Include code if helpful.]
## Attachments
[Screenshot or Loom video showing the impact — Intigriti triagers prefer video for complex bugs]
Intigriti-specific notes:
- Title format:
[Bug Class]: [One-line impact] (no formula required, but keep it specific)
- Severity is set by you: Critical/High/Medium/Low/Exceptional
- CVSS 3.1 is standard (CVSS 4.0 also accepted on newer programs)
- PoC video is valued much more than screenshot alone — record with Loom
- Safe harbor: Intigriti enforces it, be comfortable going slightly aggressive with testing
IMMUNEFI REPORT TEMPLATE
# [Bug Class] — [Protocol Name] — [Severity]
## Summary
[One paragraph with: root cause, affected function, economic impact, attack cost.
Include numbers where possible: "attacker can drain $X in Y transactions."]
## Vulnerability Details
**Contract:** `VulnerableContract.sol`
**Function:** `claimRedemption()`
**Bug Class:** Accounting State Desynchronization
**Severity:** Critical
### Root Cause
[Exact code snippet showing the vulnerable code with comments]
## Proof of Concept
```solidity
// SPDX-License-Identifier: MIT
pragma solidity ^0.8.0;
// Foundry PoC — run: forge test --match-test test_exploit -vvvv
contract ExploitTest is Test {
// ... full working exploit
}
Impact
[Quantified: "Attacker can drain X% of TVL = $Y at current rates.
Requires $Z gas. Attack is repeatable."]
Recommended Fix
[Specific code change with before/after]
---
## CVSS 3.1 QUICK SCORING
### Formula
CVSS = f(AV, AC, PR, UI, S, C, I, A)
### Metric Quick Picks
| Metric | Value | Weight | When |
|---|---|---|---|
| **Attack Vector (AV)** | Network | +0.85 | Via internet |
| | Local | +0.55 | Local access needed |
| **Attack Complexity (AC)** | Low | +0.77 | Repeatable |
| | High | +0.44 | Race/timing needed |
| **Privileges Required (PR)** | None | +0.85 | No login |
| | Low | +0.62 | Regular user account |
| | High | +0.27 | Admin account |
| **User Interaction (UI)** | None | +0.85 | No victim action |
| | Required | +0.62 | Victim must click |
| **Scope (S)** | Changed | higher | Affects browser/OS/other |
| | Unchanged | lower | Stays in app |
| **Confidentiality (C)** | High | +0.56 | All data exposed |
| | Low | +0.22 | Limited data |
| **Integrity (I)** | High | +0.56 | Can modify any data |
| **Availability (A)** | High | +0.56 | Crashes service |
### Typical Scores by Bug Class
| Bug | Typical CVSS | Severity |
|---|---|---|
| IDOR (read PII) | 6.5 | Medium |
| IDOR (write/delete) | 7.5 | High |
| Auth bypass → admin | 9.8 | Critical |
| Stored XSS (any user) | 5.4–8.8 | Med–High |
| SQLi (data exfil) | 8.6 | High |
| SSRF (cloud metadata) | 9.1 | Critical |
| Race condition (double spend) | 7.5 | High |
| GraphQL auth bypass | 8.7 | High |
| JWT none algorithm | 9.1 | Critical |
---
## SEVERITY DECISION GUIDE
### Critical (P1)
- Full account takeover of any user without interaction
- Remote code execution
- SQLi with ability to dump/modify entire DB
- Auth bypass to admin panel
- SSRF to cloud metadata → IAM credentials exfil
### High (P2)
- Mass PII exposure (email, phone, SSN, payment data)
- Privilege escalation from user to admin
- SSRF reaching internal services (data returned)
- Stored XSS executing for all users of sensitive feature
- Payment bypass / financial loss without limit
### Medium (P3)
- IDOR on specific user's non-critical data
- XSS on low-sensitivity page requiring victim interaction
- CSRF on important but non-critical action
- Rate limit bypass on OTP (with effort demonstrated)
### Low (P4)
- Information disclosure (non-sensitive, no PII)
- Clickjacking on sensitive action WITH working PoC
- CORS on limited data
---
## SEVERITY SELF-ASSESSMENT
Each YES raises severity:
- Exposes PII / health / financial data of other users? → +1 severity
- Allows account takeover or privilege escalation? → +2 severity
- Requires ZERO user interaction from victim? → +1 severity
- Affects ALL users (not specific condition)? → +1 severity
- Remotely exploitable with no internal network access? → baseline for High+
---
## DOWNGRADE COUNTERS
| Program Says | Counter With |
|---|---|
| "Requires authentication" | "Attacker needs only a free account — no special role or permission" |
| "Limited impact" | "Affects [N] users / exposes [PII type] / $[amount] at risk" |
| "Already known" | "Show me the report number — I searched hacktivity and found none" |
| "By design" | "Show me the documentation stating this is intended behavior" |
| "Low CVSS" | "CVSS doesn't capture business impact — attacker can extract [X] in [Y] minutes" |
| "Not exploitable" | "Here is the exact response showing victim's data returned to attacker session" |
---
## THE 60-SECOND PRE-SUBMIT CHECKLIST
[ ] Title follows formula: [Class] in [endpoint] allows [actor] to [impact]
[ ] First sentence states exact impact in plain English
[ ] Steps to Reproduce has exact HTTP request (copy-paste ready)
[ ] Response showing the bug is included (screenshot or JSON body)
[ ] Two test accounts used — not just one account testing itself
[ ] CVSS score calculated and included
[ ] Recommended fix is 1-2 sentences (not a lecture)
[ ] No typos in endpoint paths or parameter names
[ ] Report is < 600 words — triagers skim long reports
[ ] Severity claimed matches impact described — don't overclaim
[ ] Never used "could potentially" or "may allow"
[ ] PoC is reproducible by triager from a fresh state
---
## CVSS 4.0 QUICK REFERENCE (newer programs)
CVSS 4.0 replaced CVSS 3.1 in November 2023. Some newer programs require it.
### Key Differences from CVSS 3.1
| Metric | CVSS 3.1 | CVSS 4.0 |
|---|---|---|
| Attack Vector | Network/Adjacent/Local/Physical | Same |
| Attack Complexity | Low/High | Low/High |
| **NEW**: Attack Requirements | (didn't exist) | None/Present (replaces some PR/UI) |
| Privileges Required | None/Low/High | Same |
| User Interaction | None/Required | None/Passive/Active |
| Scope | Unchanged/Changed | REMOVED |
| **NEW**: Sub-Impact metrics | (didn't exist) | Vulnerable/Subsequent system impact |
### CVSS 4.0 Score Examples
| Finding | CVSS 4.0 Score | Vector |
|---|---|---|
| Unauthenticated RCE | 10.0 | CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H |
| IDOR read PII, auth required | 6.9 | CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N |
| Stored XSS, admin views it | 8.2 | CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:P/VC:H/VI:H/VA:N/SC:H/SI:H/SA:N |
| SSRF → cloud metadata | 8.7 | CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:H/SI:H/SA:N |
### Quick CVSS 4.0 Calculator
Use: https://www.first.org/cvss/calculator/4.0
Key fields:
VC/VI/VA = Vulnerable System Confidentiality/Integrity/Availability
SC/SI/SA = Subsequent System (downstream impact)
AT = None (no special condition) | Present (race/specific config needed)
UI = None | Passive (victim visits URL) | Active (victim takes explicit action)
**Practical rule**: If program uses CVSS 4.0 and you don't know the vector, use the calculator and include the full string starting with `CVSS:4.0/AV:...`. Programs cannot dispute a valid vector string.
---
## HUMAN TONE GUIDELINES
**Write to a person, not a system:**
- Triagers are tired. Get to the impact in sentence 1.
- Use "I" not "the researcher" — you found it, own it
- Short paragraphs, bullet points for steps
- Hyperlink relevant docs if needed
**Escalation language (when payout is being downgraded):**
"This vulnerability does not require any special privileges — only a free account."
"The exposed data includes [PII type], which is subject to GDPR requirements."
"An attacker can automate this with a simple loop — all [N] records in minutes."
"This is exploitable externally without network access to any internal system."
"The impact is equivalent to a full data breach of [feature/data type]."
**Avoid:**
- Jargon the triager might not know
- 5-paragraph explanations of what IDOR is (they know)
- Theoretical chains ("could be combined with X to...")
- Passive voice ("it was observed that...")
- Qualifying language ("seems to," "appears to")
---
## STEPS TO REPRODUCE FORMAT (triager-optimized)
```markdown
**Setup:**
- Account A (attacker): email=attacker@test.com, ID=111
- Account B (victim): email=victim@test.com, ID=222
- Both created via normal registration — no special access
**Steps:**
1. Log in as Account A
2. Send this request (replace `111` with victim ID `222`):
\```
GET /api/v2/resource/222 HTTP/1.1
Host: target.com
Authorization: Bearer ACCOUNT_A_TOKEN
\```
3. Response contains Account B's private data:
\```json
{"id": 222, "email": "victim@test.com", "name": "Victim User", "address": "..."}
\```
**Expected:** 403 Forbidden
**Actual:** 200 OK with victim's private data
Related Skills & Chains
triage-validation — When deciding whether to write a report at all. Workflow primitive: NEVER open this skill before triage-validation's 7-Question Gate passes; a finding that fails the gate should be killed, not written up.bugcrowd-reporting — When the target is a Bugcrowd program. Workflow primitive: this skill's body template is the foundation; bugcrowd-reporting overlays VRT selection, severity-request paragraph, OOS-clause rebuttals on top.evidence-hygiene — When PoC screenshots / HARs are being attached to the report. Workflow primitive: every artifact referenced in the "Supporting Materials" / "Proof of Concept" section gets routed through evidence-hygiene for cookie + PII redaction before attachment.redteam-report-template — When the engagement is an external red team (NOT bug bounty). Workflow primitive: confirm engagement mode via bb-methodology PART 0; if red-team, swap this skill out for redteam-report-template (different audience, different structure: Subject / Observations / Description / Impact / Recommendation / PoC).
Operator Notes (Claude-BugHunter)
Engagement-derived additions to the vendored foundation. Wisdom from real
authorized engagements + Phase 2 verification across this repo's 31+
skill-area live tests. The upstream methodology covers the WHAT; this
layer covers the WHEN-IT-ACTUALLY-WORKS and the FAILURE-MODES.
Title formula in practice
<asset> | <bug class> | <impact> — three components, no fluff. Triagers read titles in roughly three seconds and use them to order the queue.
- BAD: "Interesting finding on /api/users"
- BAD: "IDOR vulnerability in API"
- GOOD: "Authenticated IDOR on /api/users/{uid} -> admin email + role disclosure"
- GOOD: "Unauthenticated SSRF on /preview?url= -> AWS metadata 169.254.169.254 reachable"
The bad titles get opened last. The good titles get opened first. Same finding, different queue position, different triage day, different payout speed.
What triagers actually read
Their reading sequence on a fresh report:
- Title (3 seconds)
- First paragraph of impact / summary (15 seconds)
- The curl command or HTTP request block (30 seconds)
- Reproduction steps (only if the first three were convincing)
- Everything else (only on follow-up review)
Optimize the top of the report ruthlessly. Save narrative for the middle. Triagers who are convinced by step 3 will rubber-stamp the rest; triagers who aren't convinced by step 3 won't read step 5.
CVSS 3.1 vs Bugcrowd VRT vs H1 default severity
These three systems disagree about 30% of the time. The most common gap: a finding that scores CVSS 7.x (High) maps to Bugcrowd P4 (Low) or H1 Medium-default. When the platform default rates lower than CVSS:
- File the severity-request paragraph as the first body section. Bugcrowd respects this. (See
bugcrowd-reporting for the canonical template.)
- Anchor the request in CVSS vector string + business impact, not feelings. "CVSS 3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N = 7.5 High because Confidentiality:High applies to cross-tenant data exposure."
- Cite the platform's own VRT entry that matches your finding. Don't argue against the platform; route within it.
An authorized bug-bounty engagement saw P4-default findings escalated to P3 via the severity-request paragraph. The escalation isn't automatic — you have to ask, with grounded reasoning, in the first body section.
Evidence rotation
Everything in the submission body is logged forever by the platform. Operate accordingly:
- Use throwaway test accounts created specifically for the engagement.
- Rotate cookies / tokens after each submission (don't reuse the cookie that's pasted in the report).
- Never paste production cookies, real user emails, or real PII into the report body — redact in the PoC step.
- Screenshots of admin panels: blur the user list, blur the URL bar if it contains tokens.
Cross-link evidence-hygiene for the full capture-and-redact protocol.
Data Hygiene in Skills & Public Documentation
Skills, README files, and any public-facing documentation must be fully agnostic. Never include:
- Company names, domains, or brand names
- Employee names, emails, or personal identifiers
- Specific vulnerability details tied to identifiable targets
- Project IDs, bucket names, or internal hostnames
Use agnostic patterns instead:
BAD: "ecommerce-wine.com had PHPInfo + XMLRPC -> RCE"
GOOD: "e-commerce target had PHPInfo + XMLRPC -> RCE"
BAD: "info@tools-retailer.com exposed via REST API"
GOOD: "corporate email pattern exposed via REST API"
BAD: "target-health-saas.com MySQL 3306 OPEN"
GOOD: "healthcare SaaS target had MySQL 3306 exposed"
Per-target findings with full detail live in private output directories (e.g., /root/output/recon_us/). Skills store patterns, techniques, and methodology — never company-specific data. This applies to all SKILL.md files, README, SOUL.md, AGENTS.md, and any public-facing documentation.
When writing a skill or README:
- Use
TARGET, example.com, evil.com as placeholders
- Use sector-level descriptions (e.g., "healthcare SaaS," "e-commerce," "government portal")
- Aggregate statistics (e.g., "~7-8% of WP sites have CORS") rather than per-target specifics
- If a real-world example is essential, strip all identifying details: replace domain with sector, remove emails, anonymize project IDs
Templates by platform — when they differ
| Platform | Tone | Required Structure | Severity Mechanism |
|---|
| HackerOne | Narrative | Summary -> Steps -> Impact -> Suggested fix | Triager-set, contestable |
| Bugcrowd | Structured | Severity request -> VRT category -> Title -> Body -> Remediation | VRT-default + manual override paragraph |
| Intigriti | Between | Summary -> PoC -> Impact -> Recommendation | Researcher-proposed, triager-confirmed |
| Immunefi | PoC-first | Working PoC code -> Walkthrough -> Impact -> Severity | Foundry/Hardhat code is the primary deliverable |
Picking the wrong template style costs validity. A narrative-heavy Bugcrowd report misses the VRT mapping the triager needs; a structured H1 report reads as terse and gets follow-up questions that delay payout.
The single biggest report-writing mistake
Claiming an attack works "in theory" or "could be chained to [bigger impact]" without demonstrating it. Triage-validation Q6 (impact beyond technically possible) kills these on the validation side; report-writing has to mirror it on the writing side.
Two valid paths:
- Show concrete impact end-to-end. Capture the chain on a test account. Paste the full request/response sequence. Done.
- Downgrade the severity claim to match what you actually demonstrated. "IDOR on /api/users/{uid} reading email + role" is real and reportable; "IDOR chained to potential admin takeover" is not until you demonstrate the takeover.
Pick one. Never split the difference with "could potentially" or "may allow" — those phrases are the triager's signal that the report is theoretical, and theoretical reports get N/A.
bb-methodology — When Phase 5's report-writing step starts. Workflow primitive: Phase 5 calls /report which loads this skill for the platform-specific template (H1 / Bugcrowd / Intigriti / Immunefi).
