fix(chat): resolve both graph_id and library_agent_id in edit_agent

Fixes SECRT-1863

The edit_agent tool's schema claims it accepts either a graph ID or library
agent ID, but CoPilot often passes library_agent_id (which it naturally
grabs from context after creating/saving an agent), causing the lookup to
fail.

This adds explicit fallback logic in edit_agent to:
1. First try get_agent_as_json with the ID as-is (treats as graph_id)
2. If not found, try to resolve it as a library_agent_id
3. If found, use the library agent's graph_id to fetch the agent

This makes the schema truthful and improves robustness for CoPilot.

autogpt_platform/backend/backend/api/features/chat/tools/edit_agent.py

@@ -4,6 +4,7 @@ import logging
 from typing import Any
 
 from backend.api.features.chat.model import ChatSession
+from backend.api.features.library import db as library_db
 
 from .agent_generator import (
     AgentGeneratorNotConfiguredError,
@@ -120,6 +121,18 @@ class EditAgentTool(BaseTool):
 
         current_agent = await get_agent_as_json(agent_id, user_id)
 
+        # If not found by graph_id, try resolving as library_agent_id
+        if current_agent is None and user_id:
+            try:
+                library_agent = await library_db.get_library_agent(agent_id, user_id)
+                logger.debug(
+                    f"Resolved library_agent_id '{agent_id}' to graph_id "
+                    f"'{library_agent.graph_id}'"
+                )
+                current_agent = await get_agent_as_json(library_agent.graph_id, user_id)
+            except Exception as e:
+                logger.debug(f"Could not resolve '{agent_id}' as library_agent_id: {e}")
+
         if current_agent is None:
             return ErrorResponse(
                 message=f"Could not find agent with ID '{agent_id}' in your library.",