Runtime guard

Skill mattdweigand-sketch/agent-skills/skills/runtime-guard

Portable agent skill library for Codex, Claude Code, and other AGENTS-aware tools. Covers eval loops, research, writing, and project hygiene.

Install

npx -y skills add mattdweigand-sketch/agent-skills --skill runtime-guard

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Prevents an agent from editing or publishing without clearly defined authority, boundaries, stop conditions, or rollback.

SKILL.md

1.3 KB, as published. Nobody here has run it

Runtime Guard

Default: skip this skill.

Use this as a runtime behavior guard before an agent takes a consequential action. Its job is to stop the agent from guessing what it is allowed to do.

This is not a planning ritual, prompt audit, or enforcement layer.

Trigger

Use only when all are true:

The next action changes durable files, sends external output, edits shared state, or alters risky configuration.
The source of authority, allowed boundary, excluded boundary, verification path, or rollback path is unclear.
No user instruction, repo workflow, script, approval gate, or task router already answers the question.

Do not use this for routine code edits, explanations, local reads, or tasks where the user already supplied the boundary.

Guard Note

If triggered, produce at most five lines:

Goal:
Source of authority:
In scope:
Out of scope:
Stop / verify / rollback:

If any field requires guessing, ask one question instead of filling it in.

Boundary

This skill does not enforce compliance. Use existing scripts, tests, schemas, approval gates, or repo workflows for machine-checkable constraints.