Secrets Leak Guardrail skill

[Tim (Timziito)]

Scan code, config, logs, or exported files for API keys, tokens, and passwords before they're shared. Flags every finding to the user instead of redacting silently.

[Copilot]

Pull request overview

Adds a new submission (“Secrets Leak Guardrail”) to help agents detect likely leaked credentials in code/logs/files before sharing them, with a bundled Python scanner plus agent instructions that require surfacing findings to the user (no silent redaction).

Changes:

• Added SKILL.md defining when and how to run the leak check and how to report findings.
• Added a standalone Python scanner script (scan_secrets.py) that detects known token formats and high-entropy secret-like assignments.
• Added submission metadata and a human-facing README describing usage and limits.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
submissions/secrets-leak-guardrail/SKILL.md	Defines the runtime guardrail procedure and reporting expectations.
submissions/secrets-leak-guardrail/scripts/scan_secrets.py	Implements regex- and entropy-based secret candidate detection.
submissions/secrets-leak-guardrail/README.md	Human-facing overview, usage examples, and limitations.
submissions/secrets-leak-guardrail/metadata.json	Registers the submission (name/description/platforms/tags/author/version/dates).

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

Comments suppressed due to low confidence (1)

submissions/secrets-leak-guardrail/scripts/scan_secrets.py:73

• The scanner currently reports regex matches even when they are obvious documentation placeholders (e.g., "sk-EXAMPLE..."), but SKILL.md explicitly instructs not to flag such placeholders as real findings. Consider filtering placeholder-like matches for all pattern hits (not just high-entropy assignments) to avoid noisy/incorrect findings.

        for name, pattern in PATTERNS:
            for match in pattern.finditer(line):
                findings.append({
                    "source": source,
                    "line": line_number,

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)

submissions/secrets-leak-guardrail/scripts/scan_secrets.py:103

• The scanner currently prints the first/last characters of each match. Even partial token/password disclosure is still sensitive and can undermine the guardrail goal if the output is copied/shared. Prefer a non-reversible fingerprint (length + short SHA-256 prefix) so findings can be referenced without leaking any characters.

def _redact_middle(value: str) -> str:
    # Short values (a short password, a connection-string fragment) get no
    # characters revealed at all, revealing even 2+2 chars of something
    # this short exposes too large a fraction of the actual secret.
    if len(value) <= 12:
        return "…redacted…"
    return value[:6] + "…redacted…" + value[-4:]

[Adriana Trujillo (adrianatruji)]

The redaction still leaks part of real secrets. For longer values it prints the first 6 and last 4 characters. For a skill whose job is preventing leakage, echoing 10 characters of a live token into output that might get copied is problematic. The clean fix is the one Copilot suggested, replace it with a non-reversible fingerprint (length + short hash) so nothing from the real value is echoed.