AGENT SKILL

global-agent-guardrails

Install, audit, test, or update the Development System's global destructive-command guard for Codex, T3 Code, and Factory. Use when configuring agent safety hooks or investigating whether dangerous shell operations are blocked.

License 许可
MIT
最近更新
2026-11-14
Stars
★ 0
Forks
0

01 前言 Front Matter

name
global-agent-guardrails
description
Install, audit, test, or update the Development System's global destructive-command guard for Codex, T3 Code, and Factory. Use when configuring agent safety hooks or investigating whether dangerous shell operations are blocked.

02 资源文件 3

03 SKILL.md 正文

Global Agent Guardrails

This skill is the operator guide for the executable destructive-command policy. The hook is the enforcement layer; instructions alone are not protection.

Contract

  • Codex and T3 Code share the Codex PreToolUse adapter.
  • Factory uses its Execute PreToolUse adapter. This remains compatible with the older local Droid CLI that does not expose Factory App's newer commandBlocklist setting.
  • Existing hook and Factory settings are merged, never replaced.
  • The guard fails closed when a matched shell tool has malformed or missing command input.
  • Hard blocks do not imply sandboxing and do not replace repository permissions, review, backups, or explicit authorization.
  • The agent cannot bypass a block. A human may run the command outside the harness or deliberately remove the guard after reviewing the exact target and recovery plan.

Operate

Use the Development System CLI from its canonical checkout:

./bin/development-system guardrails-enable
./bin/development-system guardrails-audit
./bin/development-system guardrails-rollback

Use --home <isolated-home> in tests. Enabling requires the catalogued global-agent-guardrails skill to be installed for both Codex and Factory first.

The policy hard-blocks recursive forced deletion, catastrophic disk operations, destructive Git history/worktree operations, forced pushes, repository deletion, high-impact infrastructure destruction, and downloaded-code-to-shell pipelines. It intentionally does not block normal reads, ordinary file edits, dependency installation, or non-recursive deletion of a named file.

Verify

Audit must prove all of the following:

  • exact managed hook entries exist alongside pre-existing entries;
  • both installed policy engines match the catalogued skill bytes;
  • a harmless command is allowed;
  • representative dangerous commands are blocked by both adapters;
  • rollback restores exact prior configuration bytes.

Do not claim T3 Code has an independent hook runtime: it is a Codex client surface and inherits the Codex adapter.

04 同仓库更多 Skills