/fort-harden
Audit with fort, then fix the safe issues one at a time with your approval
$ npx -y skills add djadmin/fort --agent claude-codeHow it fires
How this command gets triggered: by you, by Claude, or both.
- Fires itselfClaude auto-loads it when your prompt matches the work.
- You can call itInvoke it directly when you want it.
- Slash command
/fort-harden
Context preview
What this command does when you run it.
Audit with fort, then fix the safe issues one at a time with your approval
Command definition
fort-harden.mddescription: Audit with fort, then fix the safe issues one at a time with your approval
argument-hint: "[check ids to focus on, optional]"
tools:
allowed: ["Bash", "Read"]
Audit this Mac and guide the user through fixing what's safe — **with explicit approval before any change**. This command can modify system settings, so move carefully.
1. Audit first: `fort --json` (or `fort --only $ARGUMENTS --json` if IDs were given). 2. Identify the **fixable** failing checks (`fixable: true`, `status: fail`). Present them as a short list, each with a one-line reason it matters and what fixing it would affect. Leave judgment-call items (admin rights, antivirus, OS upgrade) out of the fix list — explain those and let the user decide. 3. For each fix the user is interested in, run `fort --only <id> --dry-run` and show the exact command fort would run. Let them see it before it happens. 4. Get an explicit yes for the specific checks. Then apply exactly those: `fort --only <approved-ids> --fix` (sudo may prompt for a password — that's expected). **Never expand beyond what they approved, and never fix everything implicitly.** 5. Re-run `fort --only <ids> --json` to confirm each change took effect, and report what changed. 6. Respect the user's setup: if they say they use SSH, need an admin account, etc., do not push to change it. Their workflow wins over a generic checklist.
Use `reference/checks.md` from the fort skill for the trade-offs of each fix.
$ARGUMENTS
Read more
description: Audit with fort, then fix the safe issues one at a time with your approval argument-hint: "[check ids to focus on, optional]" tools: allowed: ["Bash", "Read"]
Audit this Mac and guide the user through fixing what's safe — **with explicit approval before any change**. This command can modify system settings, so move carefully.
1. Audit first: `fort --json` (or `fort --only $ARGUMENTS --json` if IDs were given). 2. Identify the **fixable** failing checks (`fixable: true`, `status: fail`). Present them as a short list, each with a one-line reason it matters and what fixing it would affect. Leave judgment-call items (admin rights, antivirus, OS upgrade) out of the fix list — explain those and let the user decide. 3. For each fix the user is interested in, run `fort --only <id> --dry-run` and show the exact command fort would run. Let them see it before it happens. 4. Get an explicit yes for the specific checks. Then apply exactly those: `fort --only <approved-ids> --fix` (sudo may prompt for a password — that's expected). **Never expand beyond what they approved, and never fix everything implicitly.** 5. Re-run `fort --only <ids> --json` to confirm each change took effect, and report what changed. 6. Respect the user's setup: if they say they use SSH, need an admin account, etc., do not push to change it. Their workflow wins over a generic checklist.
Use `reference/checks.md` from the fort skill for the trade-offs of each fix.
$ARGUMENTS
Know your Mac's security posture, fix the gaps, and keep it locked down. One command. fort runs 15+ security checks on your Mac, fixes what it safely can, and writes an auditor-ready report. No agent, no signup, no MDM enrollment. Just a single binary.