bridge: sanitize outbound /proc and env leaks

[benvinegar]

Summary

• add broker-side outbound sanitizer for Slack replies
• block high-risk /proc leak shapes (e.g. /proc/*/environ, /proc/*/cmdline, NUL-delimited env dumps)
• redact common token/env secret patterns before outbound send
• apply sanitization to both /send and /reply endpoints
• add unit tests for new sanitizer behavior

Validation

• node --test slack-bridge/security.test.mjs

[greptile-apps]

Greptile Summary

Adds broker-side outbound message sanitization to prevent sensitive data leaks through Slack. The implementation blocks high-risk patterns (like /proc/self/environ paths and NUL-delimited env dumps) completely and redacts known token formats (Slack, GitHub, OpenAI, AWS tokens) before sending messages through the /send and /reply endpoints.

Key changes:

• New sanitizeOutboundText function with blocking patterns for /proc paths and env dumps
• Redaction patterns for common API tokens and sensitive env assignments
• Integration into both /send and /reply endpoints with contextual logging
• Comprehensive test coverage for block, redact, and pass-through scenarios

Minor improvement opportunity:

• Regex in line 178 has redundant character class that could be simplified

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• Security-critical functionality has comprehensive test coverage, follows existing code patterns, and properly integrates into both outbound endpoints. The implementation is defensive and uses well-tested regex patterns. Only minor style suggestion identified.
• No files require special attention

Important Files Changed

Filename	Overview
slack-bridge/security.mjs	Adds sanitizeOutboundText function with blocking and redaction patterns for sensitive data leaks
slack-bridge/broker-bridge.mjs	Integrates sanitizeOutboundMessage wrapper into /send and /reply endpoints with proper logging
slack-bridge/security.test.mjs	Comprehensive test coverage for blocking, redaction, and pass-through scenarios

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Agent sends message to /send or /reply] --> B[sanitizeOutboundMessage wrapper]
    B --> C[sanitizeOutboundText function]
    C --> D{Check BLOCK patterns}
    D -->|Match found| E[Return BLOCK_FALLBACK message]
    D -->|No match| F{Check REDACT patterns}
    F -->|Matches found| G[Replace with redaction tokens]
    F -->|No matches| H[Pass through unchanged]
    E --> I[Log warning with block reasons]
    G --> J[Log warning with redact reasons]
    H --> K[No logging]
    I --> L[Send sanitized text via broker]
    J --> L
    K --> L
    L --> M[Slack message posted]

Loading

_{Last reviewed commit: 13f9f35}

[greptile-apps]

_{3 files reviewed, 1 comment}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

The character class [^\s\n\r\0] is redundant since \s already includes \n and \r. This could be simplified to [^\s\0] for cleaner regex syntax.

Prompt To Fix With AI

This is a comment left during a code review.
Path: slack-bridge/security.mjs
Line: 178

Comment:
The character class `[^\s\n\r\0]` is redundant since `\s` already includes `\n` and `\r`. This could be simplified to `[^\s\0]` for cleaner regex syntax.

How can I resolve this? If you propose a fix, please make it concise.