fix(image-generation): let LLM deliver images via message tool instead of runtime media attachment

The runtime media-attachment mechanism was broken for streaming channels
(e.g. WebSocket): the _streamed flag caused _send_once to skip the final
OutboundMessage that carried generated media, so images were never delivered.

Rather than adding complex coordination between streaming and media delivery,
delegate image delivery to the LLM: after generate_image returns artifact
paths, the next_step prompt now instructs the LLM to call the message tool
with the paths in the media parameter. This works uniformly across all
channels, streaming or not.

Remove generated_media from TurnContext, _assemble_outbound, and _state_save.
Update prompts in identity.md, SKILL.md, message tool description, and
artifacts.py to reflect the new flow.

nanobot/agent/loop.py

@@ -36,14 +36,12 @@ from nanobot.session.goal_state import (
     runner_wall_llm_timeout_s,
 )
 from nanobot.session.manager import Session, SessionManager
-from nanobot.utils.artifacts import generated_image_paths_from_messages
 from nanobot.utils.document import extract_documents
 from nanobot.utils.helpers import image_placeholder_text
 from nanobot.utils.helpers import truncate_text as truncate_text_fn
 from nanobot.utils.image_generation_intent import image_generation_prompt
 from nanobot.utils.llm_runtime import LLMRuntime
 from nanobot.utils.runtime import EMPTY_FINAL_RESPONSE_MESSAGE
-from nanobot.utils.session_attachments import merge_turn_media_into_last_assistant
 from nanobot.utils.webui_turn_helpers import (
     WebuiTurnCoordinator,
     build_bus_progress_callback,
@@ -103,7 +101,6 @@ class TurnContext:
     save_skip: int = 0
 
     outbound: OutboundMessage | None = None
-    generated_media: list[str] = field(default_factory=list)
 
     on_progress: Callable[..., Awaitable[None]] | None = None
     on_stream: Callable[[str], Awaitable[None]] | None = None
@@ -1194,7 +1191,6 @@ class AgentLoop:
         all_msgs: list[dict[str, Any]],
         stop_reason: str,
         had_injections: bool,
-        generated_media: list[str],
         on_stream: Callable[[str], Awaitable[None]] | None,
         *,
         turn_latency_ms: int | None = None,
@@ -1218,7 +1214,6 @@ class AgentLoop:
             channel=msg.channel,
             chat_id=msg.chat_id,
             content=final_content,
-            media=generated_media,
             metadata=meta,
         )
 
@@ -1348,11 +1343,6 @@ class AgentLoop:
             ctx.final_content = EMPTY_FINAL_RESPONSE_MESSAGE
 
         ctx.save_skip = 1 + len(ctx.history) + (1 if ctx.user_persisted_early else 0)
-        skip_msgs = ctx.all_messages[ctx.save_skip:]
-        ctx.generated_media = generated_image_paths_from_messages(skip_msgs)
-        mt = self.tools.get("message")
-        extra = getattr(mt, "turn_delivered_media_paths", lambda: [])() if mt else []
-        merge_turn_media_into_last_assistant(ctx.all_messages, ctx.generated_media, extra)
 
         ctx.turn_latency_ms = max(0, int((time.time() - ctx.turn_wall_started_at) * 1000))
         self._save_turn(
@@ -1380,7 +1370,6 @@ class AgentLoop:
             ctx.all_messages,
             ctx.stop_reason,
             ctx.had_injections,
-            ctx.generated_media,
             ctx.on_stream,
             turn_latency_ms=ctx.turn_latency_ms,
         )

nanobot/agent/tools/message.py

@@ -140,8 +140,8 @@ class MessageTool(Tool, ContextAware):
             "Do not use this for the normal reply in the current chat: answer naturally instead. "
             "If channel/chat_id would target the current runtime conversation, do not call this tool "
             "unless the user explicitly asked you to proactively send an existing file attachment. "
-            "When generate_image creates images in the current chat, the final assistant reply "
-            "automatically attaches them; do not call message just to announce or resend them. "
+            "When generate_image creates images in the current chat, use the message tool "
+            "with the artifact paths in the media parameter to deliver the images to the user. "
             "For proactive attachment delivery, use the 'media' parameter with file paths. "
             "Do NOT use read_file to send files — that only reads content for your own analysis."
         )