agentleFS
NhanChau2409NhanChau2409/pi-workbench/skills/poc/SKILL.md

poc

Skill0 starsChanged 53 days agoMITView on GitHub

Use when the user wants a disposable Docker proof-of-concept, lab, sandbox, install trial, repository worktree, service test, or agent experiment before changing the host or main checkout. Creates, operates, materializes, and deletes labeled POC environments with standard Docker/Git commands; OrbStack is the primary macOS backend.

5 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions5 projects use a version of this.

    Word for word · 1

    1. NhanChau2409/pi-lab★ 0100% same

    Edited · 0

    None found.

    Wrote their own version · 4

    1. suboss87/FDEOps★ 940same name
    2. gemini-cli-extensions/security★ 794same name
    3. Dan-Cleary/fdestack★ 40same name
    4. pietz/skills★ 2same 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 repositoryOne other file it gives its agents.

The file

We split this skill into the rules it gives an agent

12 rules. 0 are common, 9 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If no action is clear, ask what the user wants.

    line 25
  2. Very specific:

    Run command -v docker and docker version.

    line 29
  3. Distinctive:

    Before public ports, sensitive mounts, host installation, materialization, or deletion, read safety.md.

    line 32
  4. Distinctive:

    Ask only for missing requirements.

    line 37
  5. Distinctive:

    Keep Pi as the editor/orchestrator on the host.

    line 38
  6. Distinctive:

    Prefer visible standard commands over generated automation.

    line 39
  7. Very specific:

    Run Linux/tool commands with docker exec.

    line 47
  8. Very specific:

    Use normal docker ps, docker inspect, docker logs, docker port, docker stop, and docker start.

    line 48
  9. Distinctive:

    Never mount the Docker socket, host home, SSH/cloud credential directories, or Pi state by default.

    line 52
  10. Distinctive:

    Never remove resources based on name alone.

    line 54
  11. Distinctive:

    Use host Pi tools for files and standard Docker commands for runtime operations:

    line 63
  12. Distinctive:

    Read materialize.md. Convert useful work into reviewed repository files or commits.

    line 80

“Common” is measured: kept by at least 6 in 10 of the 4 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

POC

A POC is one or more ordinary Docker containers plus a host workspace for testing a thesis or approach before committing to it. Pi stays on the host: use host tools to read/edit workspace files and docker exec to run environment-dependent commands inside containers.

This is an agent workflow skill, not a sandbox product. Keep actions transparent, reversible, and reviewable. Do not hide Docker/Git behind a custom wrapper, do not promise containment for malicious code, and do not turn a short experiment into a second project architecture.

Interpret skill arguments as one of these lifecycle actions:

create NAME [requirements]
show [NAME]
update NAME [changed requirements]
materialize NAME [target repository]
delete NAME [--delete-worktree]
delete --all

If no action is clear, ask what the user wants. Do not create a parallel CLI or wrapper around routine Docker commands.

Start every operation

  1. Run command -v docker and docker version.
  2. On macOS with OrbStack, run orbctl status and check that docker context show is orbstack.
  3. Read lifecycle.md.
  4. Before public ports, sensitive mounts, host installation, materialization, or deletion, read safety.md.

Pi mindset

  • Make the smallest useful environment for the thesis and success condition.
  • Ask only for missing requirements; choose safe defaults and state them.
  • Keep Pi as the editor/orchestrator on the host; use containers for runtime variance.
  • Prefer visible standard commands over generated automation.
  • Preserve user agency: preview destructive or exposure changes and wait for approval.
  • Materialize outcomes as files, scripts, commits, or docs that can be reviewed outside the container.

Core rules

  • One lifecycle only: create, work/test, materialize if useful, delete.
  • Pi reads and edits the host workspace directly.
  • Run Linux/tool commands with docker exec.
  • Use normal docker ps, docker inspect, docker logs, docker port, docker stop, and docker start.
  • Label every owned resource with dev.poc.managed=true, dev.poc.name=NAME, and dev.poc.kind=KIND.
  • Port forwards bind to 127.0.0.1 unless public/LAN exposure is explicitly approved.
  • Repository work uses a Git worktree by default. Direct read-write mounting of the current checkout requires approval.
  • Never mount the Docker socket, host home, SSH/cloud credential directories, or Pi state by default.
  • Preview the exact matching resources before deletion, then ask for confirmation unless the user already explicitly requested deletion.
  • Never remove resources based on name alone; require ownership labels.
  • Report the container name, workspace path, ports, and exact follow-up Docker commands.

Create

Gather only missing requirements: name, thesis/approach to test, scratch or repository workspace, base image or Docker setup, ports, mounts, environment files, architecture, and success condition. If the user asks for an "agent sandbox", clarify whether they want accidental-change isolation, dependency/runtime isolation, or hostile-code isolation; this skill handles the first two, not strong hostile-code containment. Follow lifecycle.md.

Work

Use host Pi tools for files and standard Docker commands for runtime operations:

docker exec poc-NAME COMMAND...
docker exec -it poc-NAME bash
docker logs -f poc-NAME
docker port poc-NAME

Record successful installation and tests in workspace files rather than relying on shell history.

Update

Most container settings are immutable. Read the saved manifest, preview changes, remove only the labeled container, and recreate it while preserving the workspace and explicitly persistent volumes.

Materialize

Read materialize.md. Convert useful work into reviewed repository files or commits; do not copy an opaque mutable container into production.

Delete

Follow lifecycle.md. Delete matching owned containers, per-POC images, owned volumes, and metadata. Preserve Git worktrees and branches unless explicit removal is requested and the worktree is clean. For delete --all, preview every resource carrying the managed label and ask for confirmation; never call unrestricted Docker system prune.

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.