docs: point agent_persona.py at its live-reachable sibling port

check_agent_persona() has had no live invocation path since ADR-006
(Hook-Free Execution Model, executed 2026-08-04) deregistered PreToolUse
from ~/.claude/settings.json. Its logic was ported to a new, explicitly-
invoked validate_agent_dispatch MCP tool in mcp-pre-tool-gate/server.py.
Add a sibling-implementation pointer so a future edit to either copy
prompts a check of the other, instead of the two silently diverging.

Fixes #318.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Lhhw6mrg8Co2rU4vSS6z9N

Author: piyushmakhija28

hooks/pre_tool_enforcer/policies/agent_persona.py

@@ -193,6 +193,14 @@ def check_agent_persona(tool_name, tool_input):
     """PreToolUse policy: block general-purpose subagent spawns lacking a
     real, skill-carrying persona.
 
+    SIBLING IMPLEMENTATION NOTICE: this function is not reachable from any
+    live hook (ADR-006, Hook-Free Execution Model -- PreToolUse was
+    deregistered 2026-08-04) and is kept only for its own equivalence tests.
+    The reachable backstop is the ported `validate_agent_dispatch` MCP tool
+    in mcp-pre-tool-gate/server.py (github.com/techdeveloper-org/mcp-pre-tool-gate).
+    If you change the semantics here, check whether that port needs the same
+    change (and vice versa), or the two copies will silently diverge.
+
     A general-purpose (or unset) subagent_type must carry a library agent
     persona injected as a '---persona---' YAML block at the top of its
     prompt, per the Subagent Dispatch Contract -- and that block must