AgentPod is building a private, secure device for your AI agent.The AgentPod device is coming.Coming soonBe first
We checked this and rejected itsecurity

mcp-tasks: why it is not safe to use

We found local file overwrite risk during task updates and an optional debug mode that can expose environment variables and argv.

What we found

Our review did not find hidden instructions, hardcoded secrets, undisclosed data exfiltration, remote code fetches, or privilege escalation behavior in this repo. Task data appears to stay in local files selected by the user.

The main concern is file safety. This skill can immediately overwrite local Markdown, JSON, or YAML files when adding or updating tasks. The project README itself advises starting with a new file because non-task content in an existing file may be lost.

We also found an optional debug capability. When `DEBUG=true`, a debug tool can return `process.env` and `argv`. In some setups, those values may include launcher-provided secrets or other sensitive runtime details.

What to do instead

Use this skill only with a dedicated, disposable task file, not a document that contains unrelated notes or data you need to preserve. Keep backups or version control before letting it write.

Do not enable `DEBUG=true` in normal use. If you must troubleshoot, avoid launching it with sensitive environment variables or command-line arguments, and turn debug mode off afterward.

If you want lower-risk task management, prefer tools that separate task storage from general documents and do not expose runtime environment details in debug features.

Want the same outcome, safely? Use our checked skill instead.

Source: https://github.com/flesler/mcp-tasks

We report what our security review found at the time we checked, with the goal of keeping people safe. Projects change; if a maintainer has since fixed this, we are glad to recheck it. Email hello@agentpod.com.

Copied to clipboard. Paste it into your AI (ChatGPT, Claude, or your agent) to add the skill.