How to find passwords and API keys in Confluence
By Jake Marsh · Updated
To find passwords and API keys in Confluence, install Spillproof, open Confluence settings > Spillproof, go to the Scans tab and click Scan Confluence. It reads the title and body of every page and blog post, code blocks and link targets included, plus footer and inline comments, and lists each password, private key and token it confirms on the Findings tab with the value masked. After that first scan, every new or edited page is checked within seconds of being saved.
Why Confluence search doesn't find them
Confluence search is built for words, not for shapes like AKIA followed by 16 characters or a PEM block. You'd have to guess every prefix and format, and you'd still get hits for placeholders like password: changeme. Spillproof's detectors each check a candidate before reporting it: GitHub and npm tokens have their built-in checksums verified, OpenAI keys must contain the marker every real one has, and labelled passwords have to look real, so ${PASSWORD} and changeme are skipped.
Steps
- Install Spillproof from the Atlassian Marketplace into Confluence. You need to be a Confluence admin.
- Open Confluence settings > Spillproof.
- Check the Detectors tab. Every secret detector is on by default. Under Accuracy, pick Only checksum- or prefix-verified matches (fewest alerts) if you want the shortest first list, or leave Everything, including labelled passwords that pass the strength rules to see labelled passwords as well.
- Go to the Scans tab. To start small, type a few space keys in Space or project keys to scan (leave empty for all), for example
ENG, OPS. - Click Scan Confluence. The scan runs in the background inside Atlassian. If Confluence rate-limits it, the scan pauses and resumes by itself; Cancel stops it.
- Open the Findings tab and set the type filter to Secrets (keys, tokens, passwords) and the severity filter to Critical.
- Click the page link in the Where column to see the item. The row also says which part of it holds the value, such as the body or a comment, and how many times it appears.
What counts as a secret
| Detector | Severity | Example of what it catches |
|---|---|---|
| Private key | Critical | An RSA, EC, OpenSSH or PGP private key block with key material after the header |
| AWS access key | Critical | AKIA... key IDs, and the 40-character secret when it's nearby or labelled |
| Database URL with password | Critical | postgres://app:...@db.internal/prod |
| Stripe live key | Critical | sk_live_ and rk_live_ keys; test keys are ignored by default |
| GitHub, GitLab and Atlassian tokens | Critical | ghp_, github_pat_, glpat-, ATATT3... |
| OpenAI or Anthropic API key | Critical | Keys with OpenAI's marker, sk-ant-api03- |
| Azure storage or connection string | Critical | Account keys and shared access keys inside connection strings |
| Slack token or webhook | High | xoxb- tokens and incoming webhook URLs |
| Google API key or service account | High | AIza... keys and service-account JSON with a private key |
| Package registry, email and SMS keys | High | npm, PyPI, Docker Hub, NuGet, SendGrid, Mailgun, Twilio |
| Password or secret in plain text | High | password: ..., client_secret=... with a real-looking value |
| JSON Web Token | Medium | Signed JWTs whose header decodes, often copied from browser tools |
Where secrets tend to hide
- Runbooks and "how to connect" pages, inside code blocks.
- Incident pages, where someone pasted a log line with a token in it.
- Inline comments on old pages, which few people read again.
- Links, where a webhook URL or a URL with a password sits in the link target, not the visible text.
After you find one
Click Redact to replace the value in the current page with [redacted <type>], or Restrict page to limit the page to you and its author while you sort it out. Then rotate the key or password at its source and click Rotated: earlier page versions still hold the old text. Removing a leaked password from a page and its history covers that step by step.
What it costs
Spillproof is $5 a month for up to 10 users or $50 a year, and $1.75 per user a month from 11 users, with lower rates on larger sites. Atlassian bills it, with a 30-day free trial, and one subscription covers Confluence and Jira. Install it from the Spillproof page.
Spillproof: PII and Secret Scanner for Confluence and Jira
Find passwords, API keys, card numbers and ID numbers in Confluence and Jira. Checksum-verified, masked, nothing leaves Atlassian. From $5 a month for up to 10 users.
Questions
Does it scan page history or attachments?
No. It scans the current title, body and comments of pages and blog posts. Attachments, whiteboards, databases, templates and older page versions aren't read.
Can I scan one space first?
Yes. Type its key in Space or project keys to scan (leave empty for all) on the Scans tab before you click Scan Confluence.
Will it send our pages anywhere to check them?
No. Detection runs inside Atlassian's Forge platform, and Spillproof keeps only a masked preview and a fingerprint of each value.
What if a finding is a fake key used in documentation?
Click Not sensitive. That exact value is ignored everywhere from then on, and you can undo it under Ignored values.
More Spillproof guides
- How to find credit card numbers and Social Security numbers in Confluence and Jira
- How to scan Jira issues and comments for secrets
- How to remove a leaked password from a Confluence page and its history
More Confluence guides
- How to export Jira issues to a Confluence page as a static table (Stillframe)
- How to set a reminder on a Jira issue (Remindwell)
- How to baseline requirements in Jira (Tracestitch)
- How to find duplicate issues in Jira (Mergewell)