Skip to content
Skills Library
DevOps & cloud · Tool integration

protocol-connect

/protocol-connect

Use when the harness must reach a computer the human owns, check the route, set up a workspace there with the human's approval, verify it and undo it.

Licence
Apache-2.0
Last repository change
04.11.2026

Assessment and security

85/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content20 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
2
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 04.11.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What it does and when to use it

Use when the harness must reach a computer the human owns, check the route, set up a workspace there with the human's approval, verify it and undo it.

Original author description

Use when the harness must reach a computer the human owns, check the route, set up a workspace there with the human's approval, verify it and undo it.

How it works

  1. route
  2. identify
  3. plan
  4. research and generate
  5. deploy, verify, task, undo
Original SKILL.md headings · 22 lines, 346 words
SKILL.md preview
---
name: protocol-connect
description: Use when the harness must reach a computer the human owns, check the route, set up a workspace there with the human's approval, verify it and undo it.
---

# Reaching a computer and setting up a workspace there

Reaching another computer is the most consequential thing this harness does, so the policy lives in code and not in this text: the target comes from `targets.yaml`, which only the human edits; every write shows its exact command and its undo and waits for the human; the commands that may run are the ones in a recipe whose hash the human approved.

What this text adds is judgment, state by state.

## route
Check that the computer answers over the route its entry allows. If it does not, stop and say what is missing.

## identify
Find the operating system family. The probe skill knows the Unix-like ones. For anything else, read the computer with `remote_read` (each command is shown to the owner and approved; its output is stored by hash and you never see it) and let a skill you write turn those captures into facts.

## plan
Ask for the recipe and the candidate pairs. If no skill has a recipe for this computer, that is the unknown ladder: research, then write one. If there is a recipe but no pair fits, ask the owner: the way in is a proposal for a new runtime source, which only the owner adds.

## research and generate
Read the official documentation the owner put in the documentation folder, then write one skill: probes (read-only commands), a script that turns their output into facts, tests made from the REAL captures, a signature on the operating system fact, and a recipe. The recipe lists every write with its undo, all under one folder, nothing that fetches from the network, changes accounts, services or the firewall beyond that folder, or runs privileged. The owner approves the exact code once, by its hash.

## deploy, verify, task, undo
Choose one pair from the candidates and deploy it; every command asks again. Verification is by code (a fresh token, a real generation, loopback-only, speed and memory). One small task runs through the workspace. Then undo everything and prove nothing is left.
Files in this package (2)
  • SKILL.md 2 KB
  • skill.meta.json 6 KB

Versions

The database has no other skills named protocol-connect.

Database snapshot: 13.11.2026 · package version ac082a8ae722 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 81workspace-provisioning
    Use when a model server has to be put on another device the human owns, run there, checked and removed again.
  • 85protocol-undo
    Use when the owner asks to undo what was set up on a machine, later and in a new run. It undoes only what the recorded setup plan wrote, one step at a time, each step approved by the owner.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.