smtp-sender

A skill to send emails with support for markdown, HTML text, and attachments, leveraging existing SMTP configuration in `{skill~}/smtp-config.json`. Includes...

What to consider before installing: - The script does send email via SMTP using credentials stored in smtp-config.json next to the skill; those credentials are plaintext in that file. Consider storing SMTP credentials in a secret manager or environment variables rather than a repo file. - SKILL.md promises markdown-to-HTML conversion, retry logic, and logging — none of these are implemented. If you need those features, review and update the code (add a markdown library, add retry/error-handling and persistent logging) before relying on it. - The script will attach any file paths you supply. If you allow automatic or autonomous invocation, be careful: an agent could be instructed to attach and send arbitrary local files (risk of data exfiltration). Restrict autonomous use or add path whitelisting and validation. - The example config contains placeholder credentials and an internal IP — treat it as an example only. Ensure file permissions (chmod 600) and that the skill directory is not world-readable. - If you want to use this skill safely: audit the code, implement missing features properly (markdown conversion, retries, logging), move credentials to a secret store or environment variables, add explicit input validation for attachment paths, and restrict autonomous invocation or require explicit user confirmation before sending. Confidence note: high — the code is small and the mismatches between documentation and implementation are clear. There are no signs of network C2 endpoints or obfuscated code, but the feature/documentation inconsistencies and plaintext credential handling are noteworthy.

Type: OpenClaw Skill Name: smtp-sender Version: 1.0.2 The `email_sender.py` script is vulnerable to Local File Inclusion (LFI) via its `--body-file` and `--attachments` arguments. An AI agent, if subjected to prompt injection, could be instructed to provide paths to sensitive local files (e.g., credentials, system configuration) for these arguments, leading to their unauthorized reading and exfiltration via email. While the code itself does not exhibit explicit malicious intent, this capability presents a significant security risk that could be leveraged for data theft.