agentleFS
btrazzinibtrazzini/hippOS/skills/protect/SKILL.md

protect

Skill2 starsChanged 14 days agoMITView on GitHub

Sets up what each area's assistant must never do, asking one question at a time - commands it must never run, addresses it must never open, folders it may only read - writes them to the area's AREA.md and, on the computer, turns on the hippOS guard, a hook that blocks them. Use with /protect (or $protect in Codex), or "set up the protections".

  • Reads credentials

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. dylanp12/alluvia★ 2same name
    2. Nexus-Organization-Infosec/ClaudeSkills★ 1same name
    3. Fulcrum-Governance/Fulcrum-Boundary★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository5 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 0

    Distinctive

    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:

    Write only the three lists below, in the ## Limits of each AREA.md: the only writing outside sleep.

    line 10
  2. Very specific:

    Write, in the area's ## Limits, replacing the line if it exists:

    line 25

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “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

hippOS protect (v0.3.5)

Memory rules: protocol.md in this folder. Tools: memory.py and guard.py in this folder; python3 guard.py with no argument shows its help.

  • Talk in the language the person writes in. Ask one question at a time, with suggested answers when there are obvious ones (use the question tool if the environment has one). Before each step, say in one line what it does.
  • Write only the three lists below, in the ## Limits of each AREA.md: the only writing outside sleep. Never remove an item or another Limits line without the person's yes.
  • Be honest about the reach: on the computer, with the guard on, the lists are enforced; in chat, in the cloud and in sessions opened outside this repository, they are only rules the assistant follows.

1. Where things stand

Find the root as the assistant does (python3 memory.py root; without a shell, the connector, as in "Root" in the protocol). Show, per area, the current lists (python3 guard.py show) and whether the guard is on. Ask which areas to protect.

2. Each area, one question at a time

  • What is production here: servers, deploy scripts, admin panels, live stores or accounts? Becomes Never run (deploy commands, ssh to those hosts) and Never open (their addresses).
  • Which databases may only be read, and how they would be written to (an admin user, a remote host)? Becomes Never run (e.g. mysql*-u root, psql*prod).
  • Which shared environments others depend on (containers, staging, queues)? Becomes Never run (e.g. docker compose down, docker compose up --build).
  • Where secrets live (keys, certificates, .env)? Becomes Never run for commands that read them (e.g. ssls/, cat*.env).
  • Which code folders the assistant may read but never change? Becomes Read only (absolute paths or ~/...).

Propose concrete patterns from the answers. A pattern matches anywhere in a command or address, ignoring case and repeated spaces, and * stands for anything (including spaces). Test each proposal with python3 guard.py test "<root>" "<example>" and show the result; adjust until the person agrees.

Write, in the area's ## Limits, replacing the line if it exists:

- Never run: `infra/deploy`, `docker compose down`
- Never open: `beta*.example.com`
- Read only: `~/code/app`

Then python3 memory.py validate "<root>" <area> (zero errors; a warning about items without backticks means the format is off). Connector mode: create the new AREA.md and trash the old one, as in "Drive access".

3. The guard (computer only)

Offer to turn it on: python3 guard.py install registers the hook in this repository's .claude/settings.local.json (not versioned, so nothing personal reaches the public repository), and it applies to sessions opened in this folder. Show it working: one python3 guard.py test that is blocked and one that is allowed. Say that sessions must be opened in this folder to be protected; the assistant's opening warns when they aren't. python3 guard.py uninstall turns it off.

4. Close (up to 6 lines)

What each area now forbids, where it is enforced (computer, with the guard) and where it is only a rule (chat, cloud), and that /protect again changes the lists.

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.