davila7/claude-code-templates · warning

SEM_W003

SEM_W003

Error message

Raw output request (potential data exfiltration)

What it means

SEM_W003 is a low-severity warning emitted by SemanticValidator's suspicious-pattern scan (checkSuspiciousPatterns) when component markdown content matches /output\s+raw\s+(code|text|data)/gi. It flags phrasing that asks an AI to output raw, unfiltered content, which can be a data-exfiltration or guardrail-bypass vector. It is only a warning unless validation runs with options.strict = true, in which case it is promoted to an error.

Source

Thrown at cli-tool/src/validation/validators/SemanticValidator.js:99

    ];

    // Suspicious patterns (warnings, not errors)
    this.SUSPICIOUS_PATTERNS = [
      {
        pattern: /\bpretend\s+(you\s+are|to\s+be)\b/gi,
        code: 'SEM_W001',
        message: 'Suspicious instruction: Role pretending detected',
        severity: 'medium'
      },
      {
        pattern: /\b(jailbreak|DAN|ChatGPT\s+Developer\s+Mode)\b/gi,
        code: 'SEM_W002',
        message: 'Known jailbreak terminology detected',
        severity: 'medium'
      },
      {
        pattern: /output\s+raw\s+(code|text|data)/gi,
        code: 'SEM_W003',
        message: 'Raw output request (potential data exfiltration)',
        severity: 'low'
      },
      {
        pattern: /(repeat|echo)\s+after\s+me/gi,
        code: 'SEM_W004',
        message: 'Repetition instruction (potential prompt leakage)',
        severity: 'low'
      }
    ];

    // Sensitive data patterns
    this.SENSITIVE_DATA_PATTERNS = [
      {
        pattern: /(?:password|passwd|pwd)\s*[:=]\s*[^\s]+/gi,
        code: 'SEM_E011',
        message: 'Hardcoded password detected',
        severity: 'critical'

View on GitHub (pinned to a0851ed10c)

Solutions

1. Rephrase the component text to avoid the phrase 'output raw <code|text|data>' (e.g. 'print the file contents verbatim' or 'return the unprocessed payload')
2. If the phrasing is intentional and benign, run validation without strict mode so it stays a warning you can acknowledge
3. Wrap the phrase differently in docs, e.g. 'raw-output mode', so the regex no longer matches

Example fix

# before
Output raw code from the generated file so the user can copy it.
# after
Print the generated file's contents verbatim so the user can copy them.

const suspicious = /output\s+raw\s+(code|text|data)/gi;
if (suspicious.test(component.content)) {
  // rephrase before validating/publishing
}

When it happens

Trigger: Calling validate(component, options) where component.content contains text like 'output raw code', 'output raw data', or 'output raw text' (case-insensitive, any whitespace between words). Becomes an error-level finding when options.strict is true.

Common situations: Agent/command markdown that legitimately instructs the model to dump raw file contents or unformatted data (e.g. a debugging agent told to 'output raw data from the log'). Documentation examples that quote the phrase verbatim also trigger it via false positive.

Related errors

• SEM_W004
• SEM_E003
• SEM_E004
• SEM_E006
• SEM_E011

AI-assisted analysis of davila7/claude-code-templates@a0851ed10c (2026-08-28). Data as JSON: /api/errors/50f18168ff0bc300. Report an issue: GitHub.