Skip to main content
Be Bitwise

Writing Security Advisories

Responsible Disclosure & UK Law · Zero-Day Research track

6 min read

In this module

Why Advisories Matter

A security advisory is how vulnerability information reaches the people who need to act on it: sysadmins, security teams, the wider community. A good one lets defenders assess risk, prioritise patching, and take immediate action. A sloppy one creates confusion, delays remediation, and can leave users exposed to a risk they haven't properly understood.

Writing clear, professional advisories is a real skill. Your advisory is the public record of your work — and often the document on which organisations base their patching decisions.

Structure of a Professional Advisory

A complete advisory follows a standard structure that the security community has converged on over decades of practice. Each section serves a specific purpose.

1. Title

A concise, descriptive title that identifies the vulnerability type and the affected product:

c
Buffer Overflow in ExampleLib TIFF Parser Allows Remote Code Execution

Avoid vague titles. The title should tell a system administrator at a glance whether this advisory is relevant to their systems.

2. Advisory Metadata

c
Advisory ID: EXAMPLE-2026-001
CVE ID: CVE-2026-XXXXX
CVSS v4.0 Score: 8.7 (CVSS-B)
Date Published: 2026-04-27
Last Updated: 2026-04-27
Researcher: Your Name

Include the CVE ID once assigned, the CVSS score with the notation indicating which metric groups were used, and dates for publication and any subsequent updates. A researcher normally publishes a base score (CVSS-B): the threat and environmental metrics depend on exploit activity and each reader's own deployment, so consumers add those themselves.

3. Affected Product and Versions

Be precise. Specify the exact product name, vendor, affected versions, and which versions contain the fix. That last bit is what readers will look up first.

c
Affected: ExampleLib 3.0.0 through 3.4.2
Fixed in: ExampleLib 3.4.3
Not affected: ExampleLib 2.x (code path does not exist)
Platform: All platforms (cross-platform library)

Version ranges eliminate ambiguity. An administrator reading this knows immediately whether their installation is vulnerable and what version to upgrade to.

4. Vulnerability Description

Describe the vulnerability in clear technical language. Explain what the flaw is, where it occurs in the codebase or processing logic, and what type of vulnerability it represents (buffer overflow, use-after-free, injection, etc.).

c
A heap-based buffer overflow exists in ExampleLib's TIFF image
parser (src/tiff_decode.c, function parse_ifd_entry). When
processing a TIFF file with a specially crafted IFD entry count
exceeding 65535, the parser allocates a fixed-size buffer based
on a 16-bit field but copies data using a 32-bit length value.
This allows an attacker to write beyond the allocated buffer.
Precision Over Drama

Write advisories in neutral, technical language. Avoid sensationalism and marketing-style vulnerability naming. Your audience is security professionals who need accurate information to make decisions. State the facts clearly: what is vulnerable, how it can be triggered, and what the consequences are.

5. Steps to Reproduce

Provide clear, numbered steps that allow the vendor (and later, other researchers) to reproduce the vulnerability. This is often the most valuable part of the advisory from the vendor's perspective.

c
1. Compile ExampleLib 3.4.2 with AddressSanitizer:
$ ./configure CFLAGS="-fsanitize=address"
$ make
2. Process the attached proof-of-concept file:
$ ./examplelib_convert poc.tiff output.png
3. Observe the heap-buffer-overflow report from ASan,
showing a write of 128 bytes past the end of a
4096-byte allocation in parse_ifd_entry().

Include the minimal proof-of-concept file or input that triggers the vulnerability. If the PoC is a binary file, describe its key characteristics (file size, which bytes are modified from a valid file, etc.).

6. Impact Assessment

Explain what an attacker can achieve by exploiting this vulnerability. Be specific about the practical consequences:

c
An attacker who supplies a malicious TIFF file to an application
using the affected ExampleLib versions can achieve arbitrary code
execution in the context of the process. In common deployment
scenarios (web applications processing user-uploaded images),
this enables remote code execution without authentication.

7. Suggested Mitigations

Provide actionable advice for users who cannot immediately upgrade:

c
- Upgrade to ExampleLib 3.4.3 or later (recommended).
- If upgrading is not immediately possible, disable TIFF
processing by setting EXAMPLELIB_DISABLE_TIFF=1.
- Validate uploaded image files using an independent parser
before passing them to ExampleLib.

8. Disclosure Timeline

Document the full chronology of the disclosure process. This establishes that the researcher acted responsibly and gives context on how long the vendor took to respond.

c
2026-01-15 Vulnerability discovered during fuzz testing
2026-01-17 Vendor contacted via security@example.com (PGP encrypted)
2026-01-19 Vendor acknowledged receipt
2026-02-03 Vendor confirmed the vulnerability
2026-03-28 Vendor released ExampleLib 3.4.3
2026-04-27 Advisory published (30 days after the fix, inside
the 90-day deadline plus 30-day patch window)

9. Credits

Credit all individuals who contributed to the discovery, analysis, or remediation. Credit is a core currency in the security research community.

Advisory Template

Putting it all together, a complete advisory follows this skeleton:

c
TITLE: [Vuln type] in [Product] [Component] Allows [Impact]
ADVISORY ID: [Your ID]
CVE ID: [CVE-YYYY-NNNNN]
CVSS v4.0: [Score] ([Notation])
DATE: [YYYY-MM-DD]
AFFECTED: [Product versions]
FIXED IN: [Patched version]
DESCRIPTION: [Technical explanation of the flaw]
REPRODUCTION: [Step-by-step instructions]
IMPACT: [What an attacker can achieve]
MITIGATION: [Workarounds and fixes]
TIMELINE: [Chronological disclosure history]
CREDIT: [Researcher names]
REFERENCES: [Links to patches, vendor advisories, CVE entry]

Coordinating Advisory Content with the Vendor

In coordinated disclosure, the researcher and vendor typically agree on advisory content before publication. This coordination ensures:

  • Accuracy. The vendor can confirm technical details, identify additional affected versions, and correct any misunderstandings about the vulnerability's scope or impact.
  • Consistent messaging. When both parties publish advisories simultaneously, users receive a coherent picture rather than conflicting information.
  • Patch availability. The advisory is published on the same day (or shortly after) the patch is released, maximising the window during which users can patch before exploitation begins.

The vendor can reasonably ask you to clarify technical details or correct factual errors. What's not reasonable: demands to suppress the advisory entirely, scrub the impact assessment, or delay indefinitely. Your advisory is your publication. Their input is collaborative, not editorial control.

Your Advisory, Your Name

The advisory you publish is a professional document that carries your name and reputation. Ensure every technical claim is accurate, every reproduction step is verified, and the impact assessment is honest - neither exaggerated nor downplayed. A single inaccurate advisory can undermine years of credibility.