Support controlled multi-install for safe AI agent integrations (#2389)

* support controlled multi-install integrations

* fix: harden multi-install integration state

* refactor: isolate integration runtime helpers

* fix: address copilot review feedback

* fix: address follow-up copilot feedback

* fix: tighten integration switch semantics

* fix: address final copilot review feedback

* fix: harden integration manifest read errors

* fix: refuse symlinked shared infra paths

* test: filter expected self-test preset warning

* test: address copilot review nits

* refactor: centralize safe shared infra writes

* fix: use no-follow writes for shared infra

* fix: keep default integration atomic on template refresh

* fix: harden shared infra error paths

* fix: preflight shared infra and future state schemas

* fix: support nested shared scripts during preflight

* test: tolerate wrapped schema error output

* fix: use safe default mode for shared text writes

* fix: use posix paths in shared skip output

* fix: share project guard for integration use

* fix: centralize spec-kit project guards

* fix: use posix project paths in cli output

* fix: harden shared manifest and upgrade refresh

src/specify_cli/integrations/base.py

@@ -87,6 +87,14 @@ class IntegrationBase(ABC):
     invoke_separator: str = "."
     """Separator used in slash-command invocations (``"."`` → ``/speckit.plan``)."""
 
+    multi_install_safe: bool = False
+    """Whether this integration is declared safe to install alongside others.
+
+    Safe integrations must use a static, unique agent root, command directory,
+    and context file. Registry tests enforce those invariants for every
+    integration that sets this flag.
+    """
+
     # -- Markers for managed context section ------------------------------
 
     CONTEXT_MARKER_START = "<!-- SPECKIT START -->"