Skip to content

Address Snyk W007: credential handling in skill instructions (skills.sh audit) #1

@wrench7

Description

@wrench7

Summary

The skills.sh Snyk audit for vibe-security reports HIGH / W007 — Insecure credential handling in the skill instructions (analyzed Mar 15, 2026).

Finding (as reported)

The audit states that the skill tells the model to scan files and report file paths, relevant lines, and “before” snippets for issues such as hardcoded keys. Without explicit redaction rules, that can cause literal secrets to be repeated in assistant output.

Suggested direction

  • Add mandatory redaction rules: never echo secrets, tokens, or full credential strings; use placeholders (e.g. REDACTED, ***) and describe location without copying values.
  • Prefer high-level descriptions of vulnerable patterns over pasting sensitive literals.
  • Optionally align wording with Snyk’s concern so future audits can re-run cleanly.

References

P.S. loving the YouTube content 🫶

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions