Freeze — advisory edit scope

Apply a current-session instruction that limits which files this agent may edit. This generated runtime does not install an enforcement hook, so the scope is advisory rather than a technical lock. Never claim that edits are blocked by the runtime.

Usage

phx-freeze
phx-freeze lib/app_web priv/repo
phx-freeze status
phx-freeze off

Treat the text after the skill invocation as follows:

Invocation	Current-session behavior
No arguments	Do not edit files; investigation and reporting remain read-only.
Path prefixes	Edit only files under the listed project-relative prefixes.
status	Report the advisory scope currently established in this conversation.
off	Clear the advisory scope for subsequent work.

Do not create .claude/.freeze. That sentinel belongs to the canonical Claude Code plugin and could affect a later Claude Code session even though this runtime cannot enforce or clear it reliably.

Iron Laws

1. Never describe this scope as enforced — it is a binding instruction for the current agent, not a runtime or security boundary.
2. Never create or modify .claude/.freeze in a generated runtime — no matching enforcement component is installed here.
3. Honor the active scope until the user clears it or the focused task ends — ask before editing outside listed prefixes.
4. Keep paths project-relative — lib/foo includes that directory and its descendants, not a sibling such as lib/foobar.

Native enforcement option

This skill remains advisory and never creates Claude's .claude/.freeze sentinel. When the generated Amp plugin is installed, use the phx: edit lock palette command for a persistent workspace lock enforced at Amp's tool.call boundary. It blocks Amp-recognized edits outside the selected scope and disables shell tools while active. Unknown third-party mutating tools remain outside Amp's file-classification helper, so this is not complete Claude-hook parity.