agentleFS
Sign inStart free
dennisonbertramdennisonbertram/claude-mac-guardian/skills/malware-scan/SKILL.md

malware-scan

Skill1 starsChanged 7 months agoMITView on GitHub

This skill should be used when the user asks to "scan for malware", "is there malware on my mac", "check for viruses", "look for suspicious processes", "malware check", or mentions infection, adware, or spyware suspicions. Performs a heuristic, signature-free sweep for common persistence and process-anomaly patterns on macOS and writes a JSON result to ~/.mac-guardian/data/.

  • Reads credentials

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    When invoked by daily-health-report, return the JSON path only.

    line 47
  2. Distinctive:

    Never quarantine, move, or delete files.

    line 51
  3. Distinctive:

    Never phone home or submit samples anywhere.

    line 52
  4. Very specific:

    Do not request sudo.

    line 53

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

malware-scan

Heuristic sweep for suspicious behaviour on a macOS machine. Not a signature-based antivirus. Reports anomalies that are commonly — but not always — associated with malware or adware.

When to invoke

Trigger on phrases like: "scan for malware", "is there malware on my mac", "check for viruses", "suspicious processes".

Critical framing

Be explicit in the summary and in the mini-report that this is a heuristic sweep — not a signed antivirus product. False positives are expected. Legitimate tools (Homebrew services, VPN clients, dev tools) often create launch agents and unsigned helper binaries.

What to check

1. Processes in ephemeral / unusual paths — ps -axo pid,user,command | awk '$3 ~ /^\/(tmp|var\/tmp|private\/tmp|var\/folders)\// {print}'. Each match is a finding with severity warn.

2. Hidden directories in $HOME that look anomalous — ls -A "$HOME" | grep -E "^\." | grep -Ev "^(\.bash|\.z|\.git|\.npm|\.cache|\.cargo|\.rustup|\.config|\.local|\.ssh|\.gnupg|\.docker|\.colima|\.claude|\.cursor|\.vscode|\.idea|\.oh-my-zsh|\.CFUserTextEncoding|\.Trash|\.DS_Store|\.viminfo|\.lesshst|\.wget-hsts|\.python_history|\.node_repl_history|\.bashrc|\.profile|\.zsh|\.nvm|\.pyenv|\.rvm|\.vim|\.tool-versions|\.yarn|\.gem|\.m2|\.gradle|\.android|\.expo|\.terraform\.d|\.aws|\.kube|\.helm|\.ansible|\.serverless|\.netrc|\.curlrc|\.subversion|\.hg|\.openjdk)$". Unmatched entries: info. Record up to 20.

3. Unsigned binaries referenced by user LaunchAgents — for each .plist in ~/Library/LaunchAgents, extract the Program or first ProgramArguments entry, then run codesign --verify --verbose=0 <path>. Non-zero exit = finding warn with the program path and launch agent path.

4. Known-bad bundle IDs — grep launch agent plists for a short deny-list: com.pcv., com.searchmine, com.MyShopCoupon, com.genieo, com.installmac, com.inkeepr, com.vsearch, com.conduit, com.cinemaplus, com.trovi. Any hit = critical.

5. Cron and at persistence — crontab -l 2>/dev/null, atq 2>/dev/null. Any non-empty output = info with contents in raw.

6. Recent writes in sensitive directories (last 7 days) — find /Library/LaunchAgents /Library/LaunchDaemons ~/Library/LaunchAgents -type f -mtime -7 2>/dev/null. Any matches: individual info findings.

7. Processes with failed codesign verification — iterate top 30 user processes via ps -axo pid,comm,command | head -30, resolve the executable, run codesign --verify where a valid path exists, and record failures as warn. Skip pseudo-paths like [kernel_task] and anything under /usr/libexec, /System/, or /sbin/.

8. DNS hijack heuristics — scutil --dns | awk '/nameserver/ {print $NF}' | sort -u. If any non-private resolver is not in a small allowlist (1.1.1.1, 1.0.0.1, 8.8.8.8, 8.8.4.4, 9.9.9.9, and RFC1918 ranges), emit info.

Output contract

Same envelope as other skills. Write to ~/.mac-guardian/data/malware-scan-<ISOdate>.json.

Set top-level severity to the highest among findings, capped at warn unless check 4 (known-bad bundle IDs) fired.

Include in summary an explicit note: e.g. "Heuristic sweep: 2 warn, 5 info. Not a substitute for a signed AV tool."

Handoff

  • On direct invocation, render and open: bash ${CLAUDE_PLUGIN_ROOT}/scripts/render-single.sh malware-scan.
  • When invoked by daily-health-report, return the JSON path only.

Safety

  • Never quarantine, move, or delete files. Report paths only.
  • Never phone home or submit samples anywhere.
  • Do not request sudo.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free