#101Guide3 min

Document vulnerabilities and corrective actions

Build a traceable record from a finding to a reviewed change and a scoped recheck.

Hackaiz team · updated 21/11/2026

Illustration: a professional files evidence folders.
Illustration: a professional files evidence folders. Illustrative photo, not a screenshot.

01Start with what was actually observed

A useful vulnerability record separates the observation, its consequences and the decision to act. In a completed report, open Findings and select the relevant item. Copy the finding title, its project and analysis reference, the rule and any location supplied. Read the evidence and confidence together: a plausible weakness is not the same as a demonstrated path to sensitive information. Record uncertainty rather than converting a hypothesis into a confirmed incident.

Use the customer-facing severity exactly as displayed: Danger, To fix or To watch. Add a short explanation specific to your application, such as an access rule that might let one signed-in customer read another customer’s records. This is an example of how to describe impact, not a claim that the check found that problem in your project. Do not put real customer records or live credentials into a broadly shared tracking entry.

  1. 1Open the relevant project and analysis.
  2. 2Select the finding and read its evidence.
  3. 3Record the analysis reference and observed behavior in your own issue register.

02Give the response an owner and a reason

Keep your action register in your existing issue tracker or controlled document. Suggested fields are the affected asset, accountable person, target date, proposed action, review decision and link back to the original finding. These are recommendations for your process, not a promise of matching fields in Hackaiz. A reader should be able to distinguish “assigned” from “changed” and “changed” from “verified”.

Write the proposed treatment in operational terms. For an overly broad access rule, specify which user should retain access and which user must be refused. If you postpone work, preserve who made that decision, why it was acceptable for now and when it must be reconsidered. Do not lower the recorded observation simply because the team lacks time to address it. Business priority and technical severity answer different questions.

    03Link the finding to the reviewed change

    For eligible code fixes, Hackaiz prepares a proposal that requires human approval before publication to GitHub. Delivery is a draft pull request, meaning a proposed set of code changes awaiting review. Your team reviews, tests, merges and deploys it; Hackaiz never merges automatically. In your register, retain the proposal reference, the draft pull request link and the human decision rather than treating a generated proposal as a completed correction.

    Record the version actually deployed and any additional manual work. A code change may not update a production access policy or revoke an exposed credential by itself. Describe those separate actions without copying the credential into the record. Analysis and fix preparation use credits; review the displayed scope before approving work. Unchecked checks are not run or billed, so their absence cannot support a statement that the corresponding risk was tested.

      04Close with a result, not an assumption

      After deployment, arrange an authorized recheck and compare the original behavior with the new evidence. For websites and APIs, domain ownership proof is required before analysis: use the DNS TXT record named _hackaiz, the HTTPS verification file at /.well-known/hackaiz-verification.txt, or ask the project’s AI to create that file. A remediation record should identify the environment checked, not merely the branch where the developer made a change.

      Preserve the new analysis reference alongside the old one. If the later analysis is partial, or the relevant check did not run, document that limit and keep the action open for verification. If nothing is found in the checks that ran, describe that bounded result rather than declaring the entire application secure. The resulting chain is observation, decision, reviewed change, deployment and recheck, with each link understandable to someone outside the implementation team.

        Frequently asked questions

        Was this article helpful?

        Related articles