evidence-hygiene is a skill for Claude to use before capturing a screenshot, exporting a HAR file or attaching any proof to a bug bounty submission. It sorts sensitive material into your own account secrets, other users’ personal data, metadata a triager finds useful and throwaway test passwords, and gives each a treatment.
It covers a cookie redaction protocol, a black-bar protocol for personal data, HAR sanitization, Burp Suite screenshot hygiene, browser console proof-of-concept patterns, capture order and steps to take after submission. The goal is evidence that convinces a triager without leaking live sessions or real people’s data.
What You Get
- A table that sorts evidence data into four categories with a treatment for each
- A cookie and header redaction protocol with redaction methods ranked by practicality
- A black-bar protocol for names, emails, phone numbers and faces in other users’ data
- HAR sanitization steps with a check that no cookie or authorization values remain
- Guidance on Burp screenshots, console proof patterns, capture order and post-submission cleanup
Common Use Cases
- Preparing screenshots for a bug bounty report that touches another user’s account data
- Exporting a HAR file for a triager without leaking session tokens
- Rotating test credentials and sessions after a report is filed
What Makes It Unique
It focuses on the narrow task of what evidence to show and what to hide, and it separates your own secrets from other people’s data with different rules for each. It also lists what to leave visible, such as trace identifiers, because those help triagers match logs.
Under The Hood
The skill is a single SKILL.md of guidance with no bundled scripts. It tells the agent how to handle screenshots, HAR files and request captures. It includes a shell function that rewrites an exported HAR file with redacted cookie and header values and a check for leftovers, so the agent may read and write files in the working folder. It also shows a browser-console fetch example for proof-of-concept requests, which sends a request from the browser to the target when run. Use it only on evidence from assets you are authorized to test.