Free tool · works now

Sensitive-Data Redaction Previewer

Inspect technical notes, transcripts, and configs for API keys, private keys, auth tokens, and credentials. Preview and export clean, air-gapped notes.

11 secrets flagged·6 in prose·5 in code fences
Input Note (Markdown, Shell, or Log)100% browser-local memory
Sanitized Preview
Detected Sensitive Entities (11)
TypeLineLocationPreview / Replacement
Email Address3:9Prose<EMAIL_ADDRESS>
IPv4 Address6:62Prose<IP_ADDRESS>
Email Address7:21Prose<EMAIL_ADDRESS>
OpenAI API Key11:24Code Fence<OPENAI_API_KEY>
Anthropic API Key12:27Code Fence<ANTHROPIC_API_KEY>
GitHub Token13:22Code Fence<GITHUB_TOKEN>
AWS Access Key ID14:27Code Fence<AWS_ACCESS_KEY_ID>
Secret / Password Assignment17:14Code Fence<SECRET_VALUE>
JSON Web Token (JWT)22:9Prose<JWT_AUTH_TOKEN>
Private Key Block25:1Prose<PRIVATE_KEY_BLOCK>
Credit Card Number29:50Prose<CREDIT_CARD_NUMBER>

The Proliferation of Unmanaged Secrets in Engineering Notes

In modern software engineering and technical research, knowledge bases act as the primary working surface for daily operations. Developers frequently copy terminal outputs, debug environment files, document REST API integrations, and summarize system architecture diagrams. In the process, high-privilege credentials inadvertently enter personal note vaults:

  • AI Provider Keys: Secret keys for OpenAI (sk-proj-...) or Anthropic (sk-ant-...) pasted during quick prototyping sessions or script troubleshooting.
  • Cloud Infrastructure Access: AWS Access Key IDs (AKIA...) or temporary session tokens stored alongside deployment instructions.
  • Source Control Tokens: GitHub Personal Access Tokens (ghp_...) or fine-grained tokens (github_pat_...) included in automation setup guides.
  • Authentication Signatures: Long JSON Web Tokens (JWT) containing encoded user IDs, scopes, and cryptographic signatures captured during API traffic inspection.
  • Asymmetric Private Keys: Multi-line RSA or OpenSSH private key blocks (-----BEGIN RSA PRIVATE KEY-----) recorded in server configuration walk-throughs.

While these notes may seem isolated on your local laptop, knowledge vaults are frequently backed up across multi-device cloud storage, pushed to private Git repositories, or pasted into team wikis. The moment a note is shared with a colleague or analyzed by a third-party cloud service, unredacted credentials turn into high-severity security incidents.

Air-Gapped Local Scrubbing vs. Remote Sanitization APIs

Enterprise data loss prevention (DLP) tools typically operate by dispatching documents to remote servers or cloud endpoints for natural language processing and regex scanning. This architecture introduces a fundamental paradox: in order to determine whether your document contains sensitive secrets, you must transmit those very secrets across the Internet to an external server.

The Sensitive-Data Redaction Previewer solves this by executing 100% in your local browser client. Powered by deterministic, zero-dependency pattern matching engines, the audit runs completely inside the browser's JavaScript memory space. Zero network requests are initiated, no analytics track your keystrokes, and closing the browser tab permanently purges all processed text. You can even run this tool completely offline without an active Internet connection.

Configurable Redaction Modes: Balancing Security and Traceability

Depending on whether you are preparing notes for a client presentation, open-source publication, or internal ticket documentation, different redaction strategies are required:

  1. Descriptive Placeholders (Recommended): Replaces detected secrets with semantic tokens such as <OPENAI_API_KEY>, <AWS_ACCESS_KEY_ID>, or <EMAIL_ADDRESS>. This preserves the structural clarity and readability of code snippets, allowing teammates to understand which credential belongs where without exposing the secret itself.
  2. Partial Masking: Preserves non-sensitive prefix and suffix characters (for example, converting sk-proj-999988887777...0000 into sk-p...0000). This provides crucial traceability during incident triage, allowing you to identify which specific key was leaked without revealing the complete usable string.
  3. Complete Removal: Completely strips the credential string from the document. Ideal for public release notes or customer-facing tutorials where secret indicators should be omitted entirely.

Code Fence Awareness: Inspecting Configuration Without Compromise

Unlike naive find-and-replace scripts that indiscriminately mangle Markdown files, this tool implements full Markdown syntax awareness. It identifies opening and closing code fence boundaries (``` and ~~~), annotates whether each secret was located within prose or code blocks, and allows engineers to selectively apply redaction rules.

Furthermore, all generated audit reports can be exported as formula-guarded CSV spreadsheets. By prefixing characters like =, +, -, and @ with single quotes, the export guarantees that opening the audit ledger in spreadsheet applications like Microsoft Excel or Google Sheets will never trigger formula injection attacks.

Related Local Note Utilities

Pair this redaction previewer with other privacy-first utilities from our suite. Inspect link topologies with the Vault Health Inspector, clean web clippings with the Rich-Text Paste Cleaner, or audit note metadata using the YAML Frontmatter Validator. Explore all utilities in our browser tools catalog.

Questions

Why should notes and transcripts be sanitized before sharing or uploading?

Engineers, researchers, and technical writers regularly paste shell outputs, configuration snippets, and meeting summaries containing active API keys, authentication bearer tokens, or internal IP addresses into local notes. When those notes are later shared with teammates, committed to public repositories, or summarized by cloud AI services, secrets can leak instantly.

How does browser-local redaction differ from cloud-based data loss prevention (DLP)?

Cloud DLP services require transmitting your unredacted documents over the Internet to a remote server for processing. If the server logs requests or suffers a breach, your secrets are exposed in transit. In contrast, this redaction previewer processes all text entirely inside your browser memory using local regular expression pattern engines. No data ever leaves your device.

Does this tool detect credentials located inside Markdown fenced code blocks?

Yes. Technical credentials frequently reside inside fenced shell scripts, JSON configurations, or environment variable assignments. The engine flags all occurrences and indicates whether each secret was found within a fenced code block, while allowing you to toggle whether code fences should be redacted.

What redaction modes are supported?

Three modes are supported: Placeholder mode replaces entities with descriptive tokens like <OPENAI_API_KEY>; Mask mode preserves non-sensitive prefix/suffix characters (such as sk-p...5678) for debugging traceability; and Remove mode deletes the secret entirely.

Does the CSV export guard against spreadsheet formula injection?

Yes. In technical notes, secrets or commands often begin with symbols like =, +, -, or @. When opened in spreadsheet applications like Microsoft Excel or Google Sheets, unquoted cells could execute arbitrary macros or formulas. This tool automatically sanitizes every CSV export field to prevent formula execution.

Keep your engineering notes and meeting records completely private

Oyma runs 100% locally on macOS with Apple Neural Engine acceleration. Capture meetings, search wikilinks, and organize Markdown notes without a single byte ever touching a remote cloud server.

tools sensitive data redaction previewer downloadDownload for Mac

No account needed. Intel Mac build