agentleFS
HeyEddi-comHeyEddi-com/heyeddi-skills/skills/heyeddi-local-first/SKILL.md

heyeddi-local-first

Skill0 starsChanged 7 days agoMITView on GitHub

ALWAYS-ON local-first gate for every agent. Prepare Terraform and config locally. terraform plan is allowed and must be summarized as a warning. classify_command.py --check exits 1 on terraform apply and other live cloud mutations. Use on any infra, Terraform, gcloud, Firebase, Kubernetes, or Google account change.

  • Sends data out

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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. Distinctive:

    Before any Terraform, OpenTofu, Pulumi, gcloud, GAM, Firebase, kubectl, Helm, aws, gsutil, bq, or Google API write, run:

    line 17
  2. Very specific:

    Run the checker on the plan command.

    line 36
  3. Very specific:

    Commit and review that diff.

    line 38

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Local first

Always on for every agent that loads HeyEddi skills (Cursor, Claude, Codex, Cloud Run, or any other host). Edit code and Terraform in the repo. Check the plan. Do not change live systems.

This is the same kind of gate as verify_setup --check. The script exits 1 on a live mutation. Do not run that command.

Hard gate

Before any Terraform, OpenTofu, Pulumi, gcloud, GAM, Firebase, kubectl, Helm, aws, gsutil, bq, or Google API write, run:

python .agents/skills/heyeddi-local-first/scripts/classify_command.py --check --command "<the exact command>"

From this hub, the script path is skills/heyeddi-local-first/scripts/classify_command.py.

ExitMeaning
0 and "action": "allow"Run it. Reads and local config only.
0 and "action": "plan"Run it. Then warn the user with what the plan would create, change, or destroy. Stop.
1 and "blocked": trueDo not run it. Edit the files instead.

A plan command is terraform plan, tofu plan, pulumi preview, or ansible-playbook --check.

Process

  1. Change the Terraform, config, or app code in the working tree.
  2. Run the checker on the plan command. Exit 0 with action: plan means run the plan.
  3. Warn the user. Quote the resource addresses the plan would add, change, or destroy.
  4. Commit and review that diff. Live apply stays in the project's deploy pipeline (Cloud Build terraform-release for HeyEddi CI) or in a terminal the human runs.

What exits 1

  • terraform / tofu apply, destroy, import, refresh, state rm, workspace delete
  • gcloud writes (delete, deploy, create, submit, IAM changes). list, describe, get, config, and auth stay allowed.
  • GAM writes, including calendar account changes. gam info and gam print stay allowed.
  • firebase deploy, kubectl writes, helm upgrade, aws writes, gsutil rm, bq queries that are not --dry_run
  • curl -X POST, PUT, PATCH, or DELETE to a Google API
  • The same commands inside bash -c, python -c, $(...), or a local .sh / .py script

firebase emulators:start is local and stays allowed.

Tools

ScriptPurpose
classify_command.py --checkExit 1 on a live mutation. Plan prints a warning and exits 0.

Related

  • @heyeddi always-on list
  • @engineering-excellence for the code change gate
  • HeyEddi CI cloudbuild.yaml step terraform-release is the reviewed apply path

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.