agentleFS
curv-institutecurv-institute/atlas/fleet/AGENTS.md

atlas / fleet

AGENTS.md0 starsChanged 4 months agoMITView on GitHub

fleet.yaml (copy from fleet.example.yaml) is the source of truth for the machines this program runs on. Consult it before doing anything on a specific node. 1. Start from the inventory - fleet.yaml has each node's recorded IP, MAC, and reachability notes. 2. For a moved/unknown IP, use authoritative sources, not scanning - the node's own console (ip -4 -br addr), the router/DHCP lease table, or the known MAC matched against ip neigh. A MAC OUI lookup identifies what a mystery…

The file

We split this AGENTS.md into the rules it gives an agent

4 rules. About 1 is common, 2 are distinctive, and 1 is very specific to this project.

  • ≈1

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use the right network for the job: management/SSH over the LAN.

    line 7
  2. Common:

    When a node is unreachable or its address is unknown, work the authoritative sources first - do not jump to raw network probing (it's slow, …

    line 11
  3. Very specific:

    Prefer non-destructive checks first (hostname, ip addr, df -h, findmnt).

    line 27
  4. Distinctive:

    Don't reboot, repartition, or change network/cluster config unless explicitly asked.

    line 28

≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md 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
AGENTS.md

Fleet operating guide (optional multi-machine layer)

fleet.yaml (copy from fleet.example.yaml) is the source of truth for the machines this program runs on. Consult it before doing anything on a specific node.

Network choice

Use the right network for the job: management/SSH over the LAN; high-volume storage and node-to-node traffic over the fast fabric. Don't push large data over the slow path when a fast address exists.

Discovery & troubleshooting discipline

When a node is unreachable or its address is unknown, work the authoritative sources first - do not jump to raw network probing (it's slow, and credential/host guessing trips safety guardrails and looks like an attack).

  1. Start from the inventory - fleet.yaml has each node's recorded IP, MAC, and reachability notes.
  2. For a moved/unknown IP, use authoritative sources, not scanning - the node's own console (ip -4 -br addr), the router/DHCP lease table, or the known MAC matched against ip neigh. A MAC OUI lookup identifies what a mystery host is (vendor) without logging in.
  3. If a host is L2-unreachable (ARP INCOMPLETE / no reply): report it and STOP. Don't escalate to subnet sweeps, port scans, or trying credentials on guessed hosts.
  4. Scan / credential-test ONLY on explicit authorization, and even then never spray credentials across guessed hosts - one named host, one stated credential, then stop.
  5. A wedged-but-pingable node (ICMP + TCP:22 OK but SSH session hangs) is usually host-side, not the network; confirm with a second path + a known-good control node before touching the network.

Change hygiene

  • Prefer non-destructive checks first (hostname, ip addr, df -h, findmnt).
  • Don't reboot, repartition, or change network/cluster config unless explicitly asked.
  • Record durable changes back into fleet.yaml.
  • Out-of-bounds work (another thread's node/experiment) is gated by /mediate, not by halting.

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.