Merge pull request #5822 from loopx-project/codex/collaboration-reply-guidance-20261007

fix(collaboration): guide receiver reports as conversation answers

Author: loopx-agent

docs/architecture/rfcs/loopx-overall-roadmap-v0.md

@@ -417,6 +417,13 @@ Chat, session, collaboration and presentation owners as R2/R3 integration work.
 Do not call a candidate catalog, spinner or queued inbox receipt a completed
 worker handoff. This checkpoint does not lower R1–R3 or G1 gates.
 
+CLI and MCP receiver inbox guidance now composes the existing shared conversation answer
+instruction, making the verbatim requester-facing use of report text explicit.
+It preserves requested depth, evidence and material failures; it adds no
+summarizer, result schema or transport rewrite. R3/S5 still require actual
+original-audience qualification: guidance alone does not establish readable
+receiver results or a completed handoff.
+
 The [ordinary workspace write checkpoint](app-conversation-and-async-inbox-v0.md#ordinary-workspace-writes-default-and-revocation-checkpoint)
 adds default project writes under the host grant, explicit read-only launch/App
 settings and actual Codex sandbox enforcement on start/resume. It reuses the typed

loopx/capabilities/manager_context/README.md

@@ -390,6 +390,13 @@ override. Scoped MCP workers use `link_work` to reuse their current Todo or opaq
 evidence IDs. An unrelated busy worker establishes no request progress, and a
 short answer needs no manufactured Todo. See the [shared collaboration boundary](../../control_plane/collaboration/README.md).
 
+CLI and MCP inbox reads reuse the shared conversation answer guidance: report text can return
+verbatim to the requester's conversation, so answer the original request at a
+proportionate depth. Routine execution bookkeeping belongs in the existing
+evidence records; requested detail and material gaps remain in the answer.
+This is authoring guidance, not a summary service or a delivery-time rewrite,
+and does not certify the quality of a receiver's result.
+
 For longer work, `--phase decision` optionally returns a meaningful intermediate
 update. A ready conclusion supersedes an unsent intermediate update. Do not send
 one notification per poll, quote private deliberation, or claim an implementation

loopx/cli_commands/manager_inbox.py

@@ -10,6 +10,7 @@
     configure_evidence_scope,
 )
 from ..control_plane.projects.registry_codec import load_project_registry
+from ..presentation.answer_instruction import collaboration_answer_instruction
 
 
 def register_manager_inbox(subparsers, add_format):
@@ -187,6 +188,7 @@ def handle_manager_inbox(args, registry_path, runtime_root):
             from ..control_plane.collaboration.peers import read_inbox
             result = read_inbox(runtime_root, registry_path, args.goal_id, args.agent_id,
                                 workspace=Path.cwd(), cursor=cursor, operation_cursor=operation_cursor)
+            result["instruction"] += " " + collaboration_answer_instruction()
             result["followthrough"] += (
                 " CLI: record assessment with manager-inbox acknowledge; optionally associate existing work "
                 "with manager-inbox link. Return audience-ready results with manager-inbox report "

loopx/collaboration_mcp.py

@@ -72,7 +72,10 @@
     request,
     require_operation_id,
 )
-from .presentation.answer_instruction import conversation_answer_instruction
+from .presentation.answer_instruction import (
+    collaboration_answer_instruction,
+    conversation_answer_instruction,
+)
 
 
 _PINNED_MODULE_LAUNCHER = (
@@ -257,7 +260,7 @@ def read_context(cursor: str | None = None) -> dict:
         Follow next_cursor for later requests. Omit cursor to start a fresh scan.
         Pages are live; reading all pages does not complete outstanding work.
         """
-        return read_inbox(
+        result = read_inbox(
             root,
             registry,
             goal_id,
@@ -266,6 +269,8 @@ def read_context(cursor: str | None = None) -> dict:
             cursor=cursor,
             caller_goal_ref=caller_goal_ref,
         )
+        result["instruction"] += " " + collaboration_answer_instruction()
+        return result
 
     @server.tool()
     def assess_request(

loopx/presentation/answer_instruction.py

@@ -18,3 +18,12 @@ def conversation_answer_instruction() -> str:
         "and exact wording; do not replace a complete answer with an ID inventory or "
         "only a file path. Never include executable HTML."
     )
+
+
+def collaboration_answer_instruction() -> str:
+    """Give CLI and MCP receivers the same requester-facing answer guidance."""
+    return (
+        "Reported result text may be returned verbatim to the requester's conversation. "
+        "Compose it as their answer, not an internal work log. "
+        + conversation_answer_instruction()
+    )

loopx/semantics/project_registry_io_manifest_v1.json

@@ -783,7 +783,7 @@
     },
     {
       "site": "loopx/cli_commands/manager_inbox.py::<module>.handle_manager_inbox::codec_read:load_project_registry#1",
-      "line": 129,
+      "line": 130,
       "column": 20,
       "kind": "codec_read",
       "api": "load_project_registry",